diff --git a/include/utils/Looper.h b/include/utils/Looper.h
index 3f00b78..cc51490 100644
--- a/include/utils/Looper.h
+++ b/include/utils/Looper.h
@@ -20,9 +20,22 @@
 #include <utils/threads.h>
 #include <utils/RefBase.h>
 #include <utils/KeyedVector.h>
+#include <utils/Timers.h>
 
 #include <android/looper.h>
 
+// Currently using poll() instead of epoll_wait() since it does a better job of meeting a
+// timeout deadline.  epoll_wait() typically causes additional delays of up to 10ms
+// beyond the requested timeout.
+//#define LOOPER_USES_EPOLL
+//#define LOOPER_STATISTICS
+
+#ifdef LOOPER_USES_EPOLL
+#include <sys/epoll.h>
+#else
+#include <sys/poll.h>
+#endif
+
 /*
  * Declare a concrete type for the NDK's looper forward declaration.
  */
@@ -190,13 +203,54 @@
 
     const bool mAllowNonCallbacks; // immutable
 
-    int mEpollFd; // immutable
     int mWakeReadPipeFd;  // immutable
     int mWakeWritePipeFd; // immutable
+    Mutex mLock;
+
+#ifdef LOOPER_USES_EPOLL
+    int mEpollFd; // immutable
 
     // Locked list of file descriptor monitoring requests.
-    Mutex mLock;
-    KeyedVector<int, Request> mRequests;
+    KeyedVector<int, Request> mRequests;  // guarded by mLock
+#else
+    // The lock guards state used to track whether there is a poll() in progress and whether
+    // there are any other threads waiting in wakeAndLock().  The condition variables
+    // are used to transfer control among these threads such that all waiters are
+    // serviced before a new poll can begin.
+    // The wakeAndLock() method increments mWaiters, wakes the poll, blocks on mAwake
+    // until mPolling becomes false, then decrements mWaiters again.
+    // The poll() method blocks on mResume until mWaiters becomes 0, then sets
+    // mPolling to true, blocks until the poll completes, then resets mPolling to false
+    // and signals mResume if there are waiters.
+    bool mPolling;      // guarded by mLock
+    uint32_t mWaiters;  // guarded by mLock
+    Condition mAwake;   // guarded by mLock
+    Condition mResume;  // guarded by mLock
+
+    Vector<struct pollfd> mRequestedFds;  // must hold mLock and mPolling must be false to modify
+    Vector<Request> mRequests;            // must hold mLock and mPolling must be false to modify
+
+    ssize_t getRequestIndexLocked(int fd);
+    void wakeAndLock();
+#endif
+
+#ifdef LOOPER_STATISTICS
+    static const int SAMPLED_WAKE_CYCLES_TO_AGGREGATE = 100;
+    static const int SAMPLED_POLLS_TO_AGGREGATE = 1000;
+
+    nsecs_t mPendingWakeTime;
+    int mPendingWakeCount;
+
+    int mSampledWakeCycles;
+    int mSampledWakeCountSum;
+    nsecs_t mSampledWakeLatencySum;
+
+    int mSampledPolls;
+    int mSampledZeroPollCount;
+    int mSampledZeroPollLatencySum;
+    int mSampledTimeoutPollCount;
+    int mSampledTimeoutPollLatencySum;
+#endif
 
     // This state is only used privately by pollOnce and does not require a lock since
     // it runs on a single thread.
@@ -204,6 +258,8 @@
     size_t mResponseIndex;
 
     int pollInner(int timeoutMillis);
+    void awoken();
+    void pushResponse(int events, const Request& request);
 
     static void initTLSKey();
     static void threadDestructor(void *st);
diff --git a/libs/utils/Looper.cpp b/libs/utils/Looper.cpp
index d2dd6eb..a5363d6 100644
--- a/libs/utils/Looper.cpp
+++ b/libs/utils/Looper.cpp
@@ -19,16 +19,17 @@
 
 #include <unistd.h>
 #include <fcntl.h>
-#include <sys/epoll.h>
 
 
 namespace android {
 
+#ifdef LOOPER_USES_EPOLL
 // Hint for number of file descriptors to be associated with the epoll instance.
 static const int EPOLL_SIZE_HINT = 8;
 
 // Maximum number of file descriptors for which to retrieve poll events each iteration.
 static const int EPOLL_MAX_EVENTS = 16;
+#endif
 
 static pthread_once_t gTLSOnce = PTHREAD_ONCE_INIT;
 static pthread_key_t gTLSKey = 0;
@@ -36,9 +37,6 @@
 Looper::Looper(bool allowNonCallbacks) :
         mAllowNonCallbacks(allowNonCallbacks),
         mResponseIndex(0) {
-    mEpollFd = epoll_create(EPOLL_SIZE_HINT);
-    LOG_ALWAYS_FATAL_IF(mEpollFd < 0, "Could not create epoll instance.  errno=%d", errno);
-
     int wakeFds[2];
     int result = pipe(wakeFds);
     LOG_ALWAYS_FATAL_IF(result != 0, "Could not create wake pipe.  errno=%d", errno);
@@ -54,6 +52,11 @@
     LOG_ALWAYS_FATAL_IF(result != 0, "Could not make wake write pipe non-blocking.  errno=%d",
             errno);
 
+#ifdef LOOPER_USES_EPOLL
+    // Allocate the epoll instance and register the wake pipe.
+    mEpollFd = epoll_create(EPOLL_SIZE_HINT);
+    LOG_ALWAYS_FATAL_IF(mEpollFd < 0, "Could not create epoll instance.  errno=%d", errno);
+
     struct epoll_event eventItem;
     memset(& eventItem, 0, sizeof(epoll_event)); // zero out unused members of data field union
     eventItem.events = EPOLLIN;
@@ -61,12 +64,45 @@
     result = epoll_ctl(mEpollFd, EPOLL_CTL_ADD, mWakeReadPipeFd, & eventItem);
     LOG_ALWAYS_FATAL_IF(result != 0, "Could not add wake read pipe to epoll instance.  errno=%d",
             errno);
+#else
+    // Add the wake pipe to the head of the request list with a null callback.
+    struct pollfd requestedFd;
+    requestedFd.fd = mWakeReadPipeFd;
+    requestedFd.events = POLLIN;
+    mRequestedFds.push(requestedFd);
+
+    Request request;
+    request.fd = mWakeReadPipeFd;
+    request.callback = NULL;
+    request.ident = 0;
+    request.data = NULL;
+    mRequests.push(request);
+
+    mPolling = false;
+    mWaiters = 0;
+#endif
+
+#ifdef LOOPER_STATISTICS
+    mPendingWakeTime = -1;
+    mPendingWakeCount = 0;
+    mSampledWakeCycles = 0;
+    mSampledWakeCountSum = 0;
+    mSampledWakeLatencySum = 0;
+
+    mSampledPolls = 0;
+    mSampledZeroPollCount = 0;
+    mSampledZeroPollLatencySum = 0;
+    mSampledTimeoutPollCount = 0;
+    mSampledTimeoutPollLatencySum = 0;
+#endif
 }
 
 Looper::~Looper() {
     close(mWakeReadPipeFd);
     close(mWakeWritePipeFd);
+#ifdef LOOPER_USES_EPOLL
     close(mEpollFd);
+#endif
 }
 
 void Looper::initTLSKey() {
@@ -157,45 +193,61 @@
 #if DEBUG_POLL_AND_WAKE
     LOGD("%p ~ pollOnce - waiting: timeoutMillis=%d", this, timeoutMillis);
 #endif
+
+    int result = ALOOPER_POLL_WAKE;
+    mResponses.clear();
+    mResponseIndex = 0;
+
+#ifdef LOOPER_STATISTICS
+    nsecs_t pollStartTime = systemTime(SYSTEM_TIME_MONOTONIC);
+#endif
+
+#ifdef LOOPER_USES_EPOLL
     struct epoll_event eventItems[EPOLL_MAX_EVENTS];
     int eventCount = epoll_wait(mEpollFd, eventItems, EPOLL_MAX_EVENTS, timeoutMillis);
+    bool acquiredLock = false;
+#else
+    // Wait for wakeAndLock() waiters to run then set mPolling to true.
+    mLock.lock();
+    while (mWaiters != 0) {
+        mResume.wait(mLock);
+    }
+    mPolling = true;
+    mLock.unlock();
+
+    size_t requestedCount = mRequestedFds.size();
+    int eventCount = poll(mRequestedFds.editArray(), requestedCount, timeoutMillis);
+#endif
+
     if (eventCount < 0) {
         if (errno == EINTR) {
-            return ALOOPER_POLL_WAKE;
+            goto Done;
         }
 
         LOGW("Poll failed with an unexpected error, errno=%d", errno);
-        return ALOOPER_POLL_ERROR;
+        result = ALOOPER_POLL_ERROR;
+        goto Done;
     }
 
     if (eventCount == 0) {
 #if DEBUG_POLL_AND_WAKE
         LOGD("%p ~ pollOnce - timeout", this);
 #endif
-        return ALOOPER_POLL_TIMEOUT;
+        result = ALOOPER_POLL_TIMEOUT;
+        goto Done;
     }
 
-    int result = ALOOPER_POLL_WAKE;
-    mResponses.clear();
-    mResponseIndex = 0;
-
 #if DEBUG_POLL_AND_WAKE
     LOGD("%p ~ pollOnce - handling events from %d fds", this, eventCount);
 #endif
-    bool acquiredLock = false;
+
+#ifdef LOOPER_USES_EPOLL
     for (int i = 0; i < eventCount; i++) {
         int fd = eventItems[i].data.fd;
         uint32_t epollEvents = eventItems[i].events;
         if (fd == mWakeReadPipeFd) {
             if (epollEvents & EPOLLIN) {
-#if DEBUG_POLL_AND_WAKE
-                LOGD("%p ~ pollOnce - awoken", this);
-#endif
-                char buffer[16];
-                ssize_t nRead;
-                do {
-                    nRead = read(mWakeReadPipeFd, buffer, sizeof(buffer));
-                } while ((nRead == -1 && errno == EINTR) || nRead == sizeof(buffer));
+                awoken();
             } else {
                 LOGW("Ignoring unexpected epoll events 0x%x on wake read pipe.", epollEvents);
             }
@@ -212,11 +264,7 @@
                 if (epollEvents & EPOLLOUT) events |= ALOOPER_EVENT_OUTPUT;
                 if (epollEvents & EPOLLERR) events |= ALOOPER_EVENT_ERROR;
                 if (epollEvents & EPOLLHUP) events |= ALOOPER_EVENT_HANGUP;
-
-                Response response;
-                response.events = events;
-                response.request = mRequests.valueAt(requestIndex);
-                mResponses.push(response);
+                pushResponse(events, mRequests.valueAt(requestIndex));
             } else {
                 LOGW("Ignoring unexpected epoll events 0x%x on fd %d that is "
                         "no longer registered.", epollEvents, fd);
@@ -226,6 +274,66 @@
     if (acquiredLock) {
         mLock.unlock();
     }
+Done: ;
+#else
+    for (size_t i = 0; i < requestedCount; i++) {
+        const struct pollfd& requestedFd = mRequestedFds.itemAt(i);
+
+        short pollEvents = requestedFd.revents;
+        if (pollEvents) {
+            if (requestedFd.fd == mWakeReadPipeFd) {
+                if (pollEvents & POLLIN) {
+                    awoken();
+                } else {
+                    LOGW("Ignoring unexpected poll events 0x%x on wake read pipe.", pollEvents);
+                }
+            } else {
+                int events = 0;
+                if (pollEvents & POLLIN) events |= ALOOPER_EVENT_INPUT;
+                if (pollEvents & POLLOUT) events |= ALOOPER_EVENT_OUTPUT;
+                if (pollEvents & POLLERR) events |= ALOOPER_EVENT_ERROR;
+                if (pollEvents & POLLHUP) events |= ALOOPER_EVENT_HANGUP;
+                if (pollEvents & POLLNVAL) events |= ALOOPER_EVENT_INVALID;
+                pushResponse(events, mRequests.itemAt(i));
+            }
+            if (--eventCount == 0) {
+                break;
+            }
+        }
+    }
+
+Done:
+    // Set mPolling to false and wake up the wakeAndLock() waiters.
+    mLock.lock();
+    mPolling = false;
+    if (mWaiters != 0) {
+        mAwake.broadcast();
+    }
+    mLock.unlock();
+#endif
+
+#ifdef LOOPER_STATISTICS
+    nsecs_t pollEndTime = systemTime(SYSTEM_TIME_MONOTONIC);
+    mSampledPolls += 1;
+    if (timeoutMillis == 0) {
+        mSampledZeroPollCount += 1;
+        mSampledZeroPollLatencySum += pollEndTime - pollStartTime;
+    } else if (timeoutMillis > 0 && result == ALOOPER_POLL_TIMEOUT) {
+        mSampledTimeoutPollCount += 1;
+        mSampledTimeoutPollLatencySum += pollEndTime - pollStartTime
+                - milliseconds_to_nanoseconds(timeoutMillis);
+    }
+    if (mSampledPolls == SAMPLED_POLLS_TO_AGGREGATE) {
+        LOGD("%p ~ poll latency statistics: %0.3fms zero timeout, %0.3fms non-zero timeout", this,
+                0.000001f * float(mSampledZeroPollLatencySum) / mSampledZeroPollCount,
+                0.000001f * float(mSampledTimeoutPollLatencySum) / mSampledTimeoutPollCount);
+        mSampledPolls = 0;
+        mSampledZeroPollCount = 0;
+        mSampledZeroPollLatencySum = 0;
+        mSampledTimeoutPollCount = 0;
+        mSampledTimeoutPollLatencySum = 0;
+    }
+#endif
 
     for (size_t i = 0; i < mResponses.size(); i++) {
         const Response& response = mResponses.itemAt(i);
@@ -278,6 +386,13 @@
     LOGD("%p ~ wake", this);
 #endif
 
+#ifdef LOOPER_STATISTICS
+    // FIXME: Possible race with awoken() but this code is for testing only and is rarely enabled.
+    if (mPendingWakeCount++ == 0) {
+        mPendingWakeTime = systemTime(SYSTEM_TIME_MONOTONIC);
+    }
+#endif
+
     ssize_t nWrite;
     do {
         nWrite = write(mWakeWritePipeFd, "W", 1);
@@ -290,23 +405,51 @@
     }
 }
 
+void Looper::awoken() {
+#if DEBUG_POLL_AND_WAKE
+    LOGD("%p ~ awoken", this);
+#endif
+
+#ifdef LOOPER_STATISTICS
+    if (mPendingWakeCount == 0) {
+        LOGD("%p ~ awoken: spurious!", this);
+    } else {
+        mSampledWakeCycles += 1;
+        mSampledWakeCountSum += mPendingWakeCount;
+        mSampledWakeLatencySum += systemTime(SYSTEM_TIME_MONOTONIC) - mPendingWakeTime;
+        mPendingWakeCount = 0;
+        mPendingWakeTime = -1;
+        if (mSampledWakeCycles == SAMPLED_WAKE_CYCLES_TO_AGGREGATE) {
+            LOGD("%p ~ wake statistics: %0.3fms wake latency, %0.3f wakes per cycle", this,
+                    0.000001f * float(mSampledWakeLatencySum) / mSampledWakeCycles,
+                    float(mSampledWakeCountSum) / mSampledWakeCycles);
+            mSampledWakeCycles = 0;
+            mSampledWakeCountSum = 0;
+            mSampledWakeLatencySum = 0;
+        }
+    }
+#endif
+
+    char buffer[16];
+    ssize_t nRead;
+    do {
+        nRead = read(mWakeReadPipeFd, buffer, sizeof(buffer));
+    } while ((nRead == -1 && errno == EINTR) || nRead == sizeof(buffer));
+}
+
+void Looper::pushResponse(int events, const Request& request) {
+    Response response;
+    response.events = events;
+    response.request = request;
+    mResponses.push(response);
+}
+
 int Looper::addFd(int fd, int ident, int events, ALooper_callbackFunc callback, void* data) {
 #if DEBUG_CALLBACKS
     LOGD("%p ~ addFd - fd=%d, ident=%d, events=0x%x, callback=%p, data=%p", this, fd, ident,
             events, callback, data);
 #endif
 
-    int epollEvents = 0;
-    if (events & ALOOPER_EVENT_INPUT) epollEvents |= EPOLLIN;
-    if (events & ALOOPER_EVENT_OUTPUT) epollEvents |= EPOLLOUT;
-    if (events & ALOOPER_EVENT_ERROR) epollEvents |= EPOLLERR;
-    if (events & ALOOPER_EVENT_HANGUP) epollEvents |= EPOLLHUP;
-
-    if (epollEvents == 0) {
-        LOGE("Invalid attempt to set a callback with no selected poll events.");
-        return -1;
-    }
-
     if (! callback) {
         if (! mAllowNonCallbacks) {
             LOGE("Invalid attempt to set NULL callback but not allowed for this looper.");
@@ -319,6 +462,11 @@
         }
     }
 
+#ifdef LOOPER_USES_EPOLL
+    int epollEvents = 0;
+    if (events & ALOOPER_EVENT_INPUT) epollEvents |= EPOLLIN;
+    if (events & ALOOPER_EVENT_OUTPUT) epollEvents |= EPOLLOUT;
+
     { // acquire lock
         AutoMutex _l(mLock);
 
@@ -350,6 +498,33 @@
             mRequests.replaceValueAt(requestIndex, request);
         }
     } // release lock
+#else
+    int pollEvents = 0;
+    if (events & ALOOPER_EVENT_INPUT) pollEvents |= POLLIN;
+    if (events & ALOOPER_EVENT_OUTPUT) pollEvents |= POLLOUT;
+
+    wakeAndLock(); // acquire lock
+
+    struct pollfd requestedFd;
+    requestedFd.fd = fd;
+    requestedFd.events = pollEvents;
+
+    Request request;
+    request.fd = fd;
+    request.ident = ident;
+    request.callback = callback;
+    request.data = data;
+    ssize_t index = getRequestIndexLocked(fd);
+    if (index < 0) {
+        mRequestedFds.push(requestedFd);
+        mRequests.push(request);
+    } else {
+        mRequestedFds.replaceAt(requestedFd, size_t(index));
+        mRequests.replaceAt(request, size_t(index));
+    }
+
+    mLock.unlock(); // release lock
+#endif
     return 1;
 }
 
@@ -358,6 +533,7 @@
     LOGD("%p ~ removeFd - fd=%d", this, fd);
 #endif
 
+#ifdef LOOPER_USES_EPOLL
     { // acquire lock
         AutoMutex _l(mLock);
         ssize_t requestIndex = mRequests.indexOfKey(fd);
@@ -372,8 +548,49 @@
         }
 
         mRequests.removeItemsAt(requestIndex);
-    } // request lock
+    } // release lock
     return 1;
+#else
+    wakeAndLock(); // acquire lock
+
+    ssize_t index = getRequestIndexLocked(fd);
+    if (index >= 0) {
+        mRequestedFds.removeAt(size_t(index));
+        mRequests.removeAt(size_t(index));
+    }
+
+    mLock.unlock(); // release lock
+    return index >= 0;
+#endif
 }
 
+#ifndef LOOPER_USES_EPOLL
+ssize_t Looper::getRequestIndexLocked(int fd) {
+    size_t requestCount = mRequestedFds.size();
+
+    for (size_t i = 0; i < requestCount; i++) {
+        if (mRequestedFds.itemAt(i).fd == fd) {
+            return i;
+        }
+    }
+
+    return -1;
+}
+
+void Looper::wakeAndLock() {
+    mLock.lock();
+
+    mWaiters += 1;
+    while (mPolling) {
+        wake();
+        mAwake.wait(mLock);
+    }
+
+    mWaiters -= 1;
+    if (mWaiters == 0) {
+        mResume.signal();
+    }
+}
+#endif
+
 } // namespace android
diff --git a/libs/utils/tests/Looper_test.cpp b/libs/utils/tests/Looper_test.cpp
index afc92f8..cea1313 100644
--- a/libs/utils/tests/Looper_test.cpp
+++ b/libs/utils/tests/Looper_test.cpp
@@ -354,14 +354,6 @@
             << "addFd should return 1 because FD was added";
 }
 
-TEST_F(LooperTest, AddFd_WhenEventsIsZero_ReturnsError) {
-    Pipe pipe;
-    int result = mLooper->addFd(pipe.receiveFd, 0, 0, NULL, NULL);
-
-    EXPECT_EQ(-1, result)
-            << "addFd should return -1 because arguments were invalid";
-}
-
 TEST_F(LooperTest, AddFd_WhenIdentIsNegativeAndCallbackIsNull_ReturnsError) {
     Pipe pipe;
     int result = mLooper->addFd(pipe.receiveFd, -1, ALOOPER_EVENT_INPUT, NULL, NULL);