diff --git a/blake3.go b/blake3.go
index 14bc907..f79c10c 100644
--- a/blake3.go
+++ b/blake3.go
@@ -168,8 +168,6 @@ func (cs *chunkState) chunkCounter() uint64 {
 	return cs.n.counter
 }
 
-// complete is a helper method that reports whether a full chunk has been
-// processed.
 func (cs *chunkState) complete() bool {
 	return cs.bytesConsumed == chunkSize
 }
@@ -244,42 +242,39 @@ func parentNode(left, right [8]uint32, key [8]uint32, flags uint32) node {
 
 // Hasher implements hash.Hash.
 type Hasher struct {
-	cs         chunkState
-	key        [8]uint32
-	chainStack [54][8]uint32 // space for 54 subtrees (2^54 * chunkSize = 2^64)
-	stackSize  int           // number of chainStack elements that are valid
-	flags      uint32
-	size       int // output size, for Sum
+	cs    chunkState
+	key   [8]uint32
+	flags uint32
+	size  int // output size, for Sum
+
+	// log(n) set of Merkle subtree roots, at most one per height.
+	stack [54][8]uint32 // 2^54 * chunkSize = 2^64
+	used  uint64        // bit vector indicating which stack elems are valid; also number of chunks added
+}
+
+func (h *Hasher) hasSubtreeAtHeight(i uint64) bool {
+	return h.used&(1<<i) != 0
 }
 
 // addChunkChainingValue appends a chunk to the right edge of the Merkle tree.
-func (h *Hasher) addChunkChainingValue(cv [8]uint32, totalChunks uint64) {
-	// This chunk might complete some subtrees. For each completed subtree, its
-	// left child will be the current top entry in the CV stack, and its right
-	// child will be the current value of cv. Pop each left child off the stack,
-	// merge it with cv, and overwrite cv with the result. After all these
-	// merges, push the final value of cv onto the stack. The number of
-	// completed subtrees is given by the number of trailing 0-bits in the new
-	// total number of chunks.
-	for totalChunks&1 == 0 {
-		// pop and merge
-		h.stackSize--
-		cv = parentNode(h.chainStack[h.stackSize], cv, h.key, h.flags).chainingValue()
-		totalChunks >>= 1
+func (h *Hasher) addChunkChainingValue(cv [8]uint32) {
+	// seek to first open stack slot, merging subtrees as we go
+	i := uint64(0)
+	for ; h.hasSubtreeAtHeight(i); i++ {
+		cv = parentNode(h.stack[i], cv, h.key, h.flags).chainingValue()
 	}
-	h.chainStack[h.stackSize] = cv
-	h.stackSize++
+	h.stack[i] = cv
+	h.used++
 }
 
 // rootNode computes the root of the Merkle tree. It does not modify the
 // chainStack.
 func (h *Hasher) rootNode() node {
-	// Starting with the node from the current chunk, compute all the
-	// parent chaining values along the right edge of the tree, until we
-	// have the root node.
 	n := h.cs.node()
-	for i := h.stackSize - 1; i >= 0; i-- {
-		n = parentNode(h.chainStack[i], n.chainingValue(), h.key, h.flags)
+	for i := uint64(bits.TrailingZeros64(h.used)); i < 64; i++ {
+		if h.hasSubtreeAtHeight(i) {
+			n = parentNode(h.stack[i], n.chainingValue(), h.key, h.flags)
+		}
 	}
 	n.flags |= flagRoot
 	return n
@@ -288,7 +283,7 @@ func (h *Hasher) rootNode() node {
 // Reset implements hash.Hash.
 func (h *Hasher) Reset() {
 	h.cs = newChunkState(h.key, 0, h.flags)
-	h.stackSize = 0
+	h.used = 0
 }
 
 // BlockSize implements hash.Hash.
@@ -306,9 +301,8 @@ func (h *Hasher) Write(p []byte) (int, error) {
 		// chunks).
 		if h.cs.complete() {
 			cv := h.cs.node().chainingValue()
-			totalChunks := h.cs.chunkCounter() + 1
-			h.addChunkChainingValue(cv, totalChunks)
-			h.cs = newChunkState(h.key, totalChunks, h.flags)
+			h.addChunkChainingValue(cv)
+			h.cs = newChunkState(h.key, h.cs.chunkCounter()+1, h.flags)
 		}
 
 		// Compress input bytes into the current chunk state.