diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index b02dadfcaf..7eda8bcdc8 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -1363,12 +1363,17 @@ typedef struct AVCodecContext {
      *   encoded input.
      *
      * Audio:
-     *   For encoding, this is the number of "priming" samples added to the
-     *   beginning of the stream. The decoded output will be delayed by this
-     *   many samples relative to the input to the encoder. Note that this
-     *   field is purely informational and does not directly affect the pts
-     *   output by the encoder, which should always be based on the actual
-     *   presentation time, including any delay.
+     *   For encoding, this is the number of "priming" samples added by the
+     *   encoder to the beginning of the stream. The decoded output will be
+     *   delayed by this many samples relative to the input to the encoder (or
+     *   more, if the decoder adds its own padding).
+     *   The timestamps on the output packets are adjusted by the encoder so
+     *   that they always refer to the first sample of the data actually
+     *   contained in the packet, including any added padding.
+     *   E.g. if the timebase is 1/samplerate and the timestamp of the first
+     *   input sample is 0, the timestamp of the first output packet will be
+     *   -delay.
+     *
      *   For decoding, this is the number of samples the decoder needs to
      *   output before the decoder's output is valid. When seeking, you should
      *   start decoding this many samples prior to your desired seek point.