diff --git a/src/compression/brotli.cpp b/src/compression/brotli.cpp
index d06563f0..1bbab824 100644
--- a/src/compression/brotli.cpp
+++ b/src/compression/brotli.cpp
@@ -113,7 +113,7 @@ class brotli_block_decompressor final : public block_decompressor::impl {
     }
     if (!::BrotliDecoderSetParameter(decoder_.get(),
                                      BROTLI_DECODER_PARAM_LARGE_WINDOW, 1)) {
-      DWARFS_THROW(runtime_error, "could not set brotli decoder paramter");
+      DWARFS_THROW(runtime_error, "could not set brotli decoder parameter");
     }
     try {
       decompressed_.reserve(uncompressed_size_);
@@ -148,7 +148,7 @@ class brotli_block_decompressor final : public block_decompressor::impl {
 
     if (res == BROTLI_DECODER_RESULT_ERROR) {
       DWARFS_THROW(runtime_error,
-                   fmt::format("brotli errro: {}", brotli_error()));
+                   fmt::format("brotli error: {}", brotli_error()));
     }
 
     decompressed_.resize(std::distance(decompressed_.data(), next_out));
diff --git a/src/reader/internal/metadata_v2.cpp b/src/reader/internal/metadata_v2.cpp
index 7f25fdb6..f75a6acd 100644
--- a/src/reader/internal/metadata_v2.cpp
+++ b/src/reader/internal/metadata_v2.cpp
@@ -1680,7 +1680,7 @@ void metadata_<LoggerPolicy>::access(inode_view iv, int mode,
                            mode, uid, gid);
 
   if (mode == F_OK) {
-    // easy; we're only interested in the file's existance
+    // easy; we're only interested in the file's existence
     ec.clear();
     return;
   }
diff --git a/src/writer/segmenter.cpp b/src/writer/segmenter.cpp
index 46df1cdd..2df0b1b8 100644
--- a/src/writer/segmenter.cpp
+++ b/src/writer/segmenter.cpp
@@ -65,7 +65,7 @@ namespace {
 /**
  * Segmenter Strategy
  *
- * For each *block*, start new rolling hash. The hashes are associcated
+ * For each *block*, start new rolling hash. The hashes are associated
  * with the block, new hash-offset-pairs will only be added as the block
  * grows. We only need to store a hash-offset-pair every N bytes, with N
  * being configurable (typically half of the window size so we find all
@@ -76,7 +76,7 @@ namespace {
  * will be scanned for matches. Old file data beyond the moving window will
  * be added to the current *block*, causing the rolling *block* hash to also
  * advance. Data needs to be added to the block only in increments at which
- * a new hash valus is computed.
+ * a new hash values is computed.
  *
  * This strategy ensures that we're using a deterministic amount of memory
  * (proportional to block size and history block count).