diff --git a/src/indexcodec.cpp b/src/indexcodec.cpp
index 47559710..8b650b4e 100644
--- a/src/indexcodec.cpp
+++ b/src/indexcodec.cpp
@@ -147,6 +147,10 @@ size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, cons
 
 	assert(index_count % 3 == 0);
 
+	// the minimum valid encoding is 1 byte per triangle and a 16-byte codeaux table
+	if (buffer_size < index_count / 3 + 16)
+		return 0;
+
 	EdgeFifo edgefifo;
 	memset(edgefifo, -1, sizeof(edgefifo));
 
@@ -161,6 +165,7 @@ size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, cons
 
 	unsigned char* code = buffer;
 	unsigned char* data = buffer + index_count / 3;
+	unsigned char* data_safe_end = buffer + buffer_size - 16;
 
 	// use static encoding table; it's possible to pack the result and then build an optimal table and repack
 	// for now we keep it simple and use the table that has been generated based on symbol frequency on a training mesh set
@@ -172,6 +177,12 @@ size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, cons
 
 	for (size_t i = 0; i < index_count; i += 3)
 	{
+		// make sure we have enough space to write a triangle
+		// each triangle writes at most 16 bytes: 1b for codeaux and 5b for each free index
+		// after this we can be sure we can write without extra bounds checks
+		if (data > data_safe_end)
+			return 0;
+
 		int fer = getEdgeFifo(edgefifo, indices[i + 0], indices[i + 1], indices[i + 2], edgefifooffset);
 
 		if (fer >= 0 && (fer >> 2) < 15)
@@ -256,6 +267,10 @@ size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, cons
 		}
 	}
 
+	// make sure we have enough space to write codeaux table
+	if (data > data_safe_end)
+		return 0;
+
 	// add codeaux encoding table to the end of the stream; this is used for decoding codeaux *and* as padding
 	// we need padding for decoding to be able to assume that each triangle is encoded as <= 16 bytes of extra data
 	// this is enough space for aux byte + 5 bytes per varint index which is the absolute worst case for any input
@@ -266,7 +281,6 @@ size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, cons
 
 	assert(data >= buffer + index_count / 3 + 16);
 	assert(data <= buffer + buffer_size);
-	(void)buffer_size;
 
 	return data - buffer;
 }
diff --git a/src/meshoptimizer.h b/src/meshoptimizer.h
index ffd3fedb..b0e2e72c 100644
--- a/src/meshoptimizer.h
+++ b/src/meshoptimizer.h
@@ -88,6 +88,7 @@ MESHOPTIMIZER_API size_t meshopt_optimizeVertexFetch(void* destination, unsigned
 /**
  * Experimental: Index buffer encoder
  * Encodes index data into an array of bytes that is generally much smaller (<1.5 bytes/triangle) and compresses better (<1 bytes/triangle) compared to original.
+ * Returns encoded data size on success, 0 on error; the only error condition is if buffer doesn't have enough space
  * For maximum efficiency the index buffer being encoded has to be optimized for vertex cache and vertex fetch first.
  *
  * buffer must contain enough space for the encoded index buffer (use meshopt_encodeIndexBufferBound to estimate)
@@ -98,7 +99,7 @@ MESHOPTIMIZER_API size_t meshopt_encodeIndexBufferBound(size_t index_count, size
 /**
  * Experimental: Index buffer decoder
  * Decodes index data from an array of bytes generated by meshopt_encodeIndexBuffer
- * Returns 0 if encoding is successful, and an error code otherwise
+ * Returns 0 if decoding was successful, and an error code otherwise
  *
  * destination must contain enough space for the resulting index buffer (index_count elements)
  */
@@ -118,6 +119,7 @@ MESHOPTIMIZER_API size_t meshopt_simplify(unsigned int* destination, const unsig
  * Experimental: Mesh stripifier
  * Converts a previously vertex cache optimized triangle list to triangle strip, stitching strips using restart index
  * Returns the number of indices in the resulting strip, with destination containing new index data
+ * For maximum efficiency the index buffer being converted has to be optimized for vertex cache first.
  *
  * destination must contain enough space for the worst case target index buffer (index_count / 3 * 4 elements)
  */