diff --git a/Changes b/Changes
index f65c43326..dca812526 100644
--- a/Changes
+++ b/Changes
@@ -6155,7 +6155,7 @@ Bug Fixes:
 - #5575: Random states are not marshallable across architectures
 - #5579: camlp4: when a plugin is loaded in the toplevel,
   Token.Filter.define_filter has no effect before the first syntax error
-- #5585: typo: "explicitely"
+- #5585: typo: "explicitly"
 - #5587: documentation: "allows to" is not correct English
 - #5593: remove C file when -output-obj fails
 - #5597: register names for instrtrace primitives in embedded bytecode
@@ -6437,7 +6437,7 @@ Native-code compiler:
   float comparisons.
 
 Standard library:
-- Format: new function ikfprintf analoguous to ifprintf with a continuation
+- Format: new function ikfprintf analogous to ifprintf with a continuation
   argument.
 * #4210, #4245: stricter range checking in string->integer conversion
   functions (int_of_string, Int32.of_string, Int64.of_string,
diff --git a/News b/News
index 94371099b..79f3f72fe 100644
--- a/News
+++ b/News
@@ -156,7 +156,7 @@ Some highlights include:
 -   Instrumentation support for fuzzing with afl-fuzz.
     (GPR#504, by Stephen Dolan)
 
--   The compilers now accept new `-args/-args0 <file>` comand-line
+-   The compilers now accept new `-args/-args0 <file>` command-line
     parameters to provide extra command-line arguments in a file.  User
     programs may implement similar options using the new `Expand`
     constructor of the `Arg` module.
diff --git a/driver/makedepend.ml b/driver/makedepend.ml
index d94940566..4942eab0f 100644
--- a/driver/makedepend.ml
+++ b/driver/makedepend.ml
@@ -578,7 +578,7 @@ let main () =
      "-all", Arg.Set all_dependencies,
         " Generate dependencies on all files";
      "-allow-approx", Arg.Set allow_approximation,
-        " Fallback to a lexer-based approximation on unparseable files";
+        " Fallback to a lexer-based approximation on unparsable files";
      "-as-map", Arg.Set Clflags.transparent_modules,
       " Omit delayed dependencies for module aliases (-no-alias-deps -w -49)";
       (* "compiler uses -no-alias-deps, and no module is coerced"; *)
diff --git a/manual/styles/html.sty b/manual/styles/html.sty
index 6a9e92535..137fdf163 100644
--- a/manual/styles/html.sty
+++ b/manual/styles/html.sty
@@ -111,7 +111,7 @@
 % Changed \next to \html@next to prevent clashes with other sty files
 % (mike@emn.fr)
 % Changed \html@next to \htmlnext so the \makeatletter and
-% \makeatother commands could be removed (they were cuasing other
+% \makeatother commands could be removed (they were causing other
 % style files - changebar.sty - to crash) (nikos@cbl.leeds.ac.uk)
 
 
diff --git a/middle_end/flambda/flambda.ml b/middle_end/flambda/flambda.ml
index 243e2e3f9..aea872f90 100644
--- a/middle_end/flambda/flambda.ml
+++ b/middle_end/flambda/flambda.ml
@@ -1138,7 +1138,7 @@ let create_set_of_closures ~function_decls ~free_vars ~specialised_args
        This would be true when the function is known never to have
        been inlined.
 
-       Note that something like that may maybe enforcable in
+       Note that something like that may maybe enforceable in
        inline_and_simplify, but there is no way to do that on other
        passes.
 
diff --git a/otherlibs/win32unix/createprocess.c b/otherlibs/win32unix/createprocess.c
index d238e53e1..758a98f92 100644
--- a/otherlibs/win32unix/createprocess.c
+++ b/otherlibs/win32unix/createprocess.c
@@ -53,7 +53,7 @@ static DWORD do_create_process_native(wchar_t * exefile, wchar_t * cmdline,
     err = GetLastError(); goto ret3;
   }
   /* If we do not have a console window, then we must create one
-     before running the process (keep it hidden for apparence).
+     before running the process (keep it hidden for appearance).
      If we are starting a GUI application, the newly created
      console should not matter. */
   if (win_has_console())
diff --git a/runtime/caml/backtrace_prim.h b/runtime/caml/backtrace_prim.h
index 7b0fd4f75..d2500cb3f 100644
--- a/runtime/caml/backtrace_prim.h
+++ b/runtime/caml/backtrace_prim.h
@@ -101,7 +101,7 @@ value caml_remove_debug_info(code_t start);
  * The first function, [caml_current_callstack_size] computes the size
  * (in words) of the needed buffer, while the second actually writes
  * the call stack to the buffer as an object of type
- * [raw_backtrace]. It should always be called with a bufer of the
+ * [raw_backtrace]. It should always be called with a buffer of the
  * size predicted by [caml_current_callstack_size]. The reason we use
  * two separated functions is to allow using either [caml_alloc] (for
  * performance) or [caml_alloc_shr] (when we need to avoid a call to
diff --git a/runtime/memprof.c b/runtime/memprof.c
index 7c5aab03a..bc87d6e29 100644
--- a/runtime/memprof.c
+++ b/runtime/memprof.c
@@ -123,7 +123,7 @@ static int32_t mt_generate_poisson(double len)
     next_mt_generate_poisson -= cur_lambda;
     if(next_mt_generate_poisson > 0) {
       /* Fast path if [cur_lambda] is small: we reuse the same
-         exponential sample accross several calls to
+         exponential sample across several calls to
          [mt_generate_poisson]. */
       return 0;
     } else {
@@ -461,7 +461,7 @@ void caml_memprof_track_young(tag_t tag, uintnat wosize)
   /* Write the ephemeron if not [None]. */
   if(Is_block(ephe)) {
     /* Subtlety: we are actually writing the ephemeron with an invalid
-       (unitialized) block. This is correct for two reasons:
+       (uninitialized) block. This is correct for two reasons:
           - The logic of [caml_ephemeron_set_key] never inspects the content of
             the block. In only checks that the block is young.
           - The allocation and initialization happens right after returning
diff --git a/runtime/signals.c b/runtime/signals.c
index 218310d4b..c6c71947c 100644
--- a/runtime/signals.c
+++ b/runtime/signals.c
@@ -177,7 +177,7 @@ CAMLexport void caml_leave_blocking_section(void)
      Another case where this is necessary (even in a single threaded
      setting) is when the blocking section unmasks a pending signal:
      If the signal is pending and masked but has already been
-     examinated by [caml_process_pending_signals], then
+     examined by [caml_process_pending_signals], then
      [caml_signals_are_pending] is 0 but the signal needs to be
      handled at this point. */
   caml_signals_are_pending = 1;
diff --git a/runtime/signals_nat.c b/runtime/signals_nat.c
index 05ed30258..42224b685 100644
--- a/runtime/signals_nat.c
+++ b/runtime/signals_nat.c
@@ -76,7 +76,7 @@ void caml_garbage_collection(void)
 {
   /* TEMPORARY: if we have just sampled an allocation in native mode,
      we simply renew the sample to ignore it. Otherwise, renewing now
-     will not have any efect on the sampling distribution, because of
+     will not have any effect on the sampling distribution, because of
      the memorylessness of the Poisson process. */
   caml_memprof_renew_minor_sample();
   if (caml_requested_major_slice || caml_requested_minor_gc ||
diff --git a/stdlib/filename.mli b/stdlib/filename.mli
index b05ad0d2e..cc018b424 100644
--- a/stdlib/filename.mli
+++ b/stdlib/filename.mli
@@ -52,7 +52,7 @@ val check_suffix : string -> string -> bool
 val chop_suffix : string -> string -> string
 (** [chop_suffix name suff] removes the suffix [suff] from
    the filename [name]. The behavior is undefined if [name] does not
-   end with the suffix [suff]. It is thus recommmended to use
+   end with the suffix [suff]. It is thus recommended to use
    [chop_suffix_opt] instead.
 *)
 
diff --git a/testsuite/tests/typing-gadts/ambiguity.ml b/testsuite/tests/typing-gadts/ambiguity.ml
index ab6c97acf..87f58c626 100644
--- a/testsuite/tests/typing-gadts/ambiguity.ml
+++ b/testsuite/tests/typing-gadts/ambiguity.ml
@@ -104,7 +104,7 @@ Error: This expression has type b = a but an expression was expected of type
    representative for an ambivalent type escaping its scope.
    The commit that was implemented poses problems of its own: we are now
    unifying the type of the patterns in the environment of each pattern, instead
-   of the outter one. The code discussed in PR#7617 passes because each branch
+   of the outer one. The code discussed in PR#7617 passes because each branch
    contains the same equation, but consider the following cases: *)
 
 let f (type a b) (x : (a, b) eq) =
diff --git a/typing/persistent_env.mli b/typing/persistent_env.mli
index 765a7b02c..d04034ef8 100644
--- a/typing/persistent_env.mli
+++ b/typing/persistent_env.mli
@@ -69,7 +69,7 @@ val check : 'a t -> (Persistent_signature.t -> 'a)
    [penv] (it may have failed) *)
 val looked_up : 'a t -> modname -> bool
 
-(* [is_imported penv md] checks if [md] has been succesfully
+(* [is_imported penv md] checks if [md] has been successfully
    imported in the environment [penv] *)
 val is_imported : 'a t -> modname -> bool
 
diff --git a/typing/typemod.ml b/typing/typemod.ml
index 93ed01eff..b081d652f 100644
--- a/typing/typemod.ml
+++ b/typing/typemod.ml
@@ -616,7 +616,7 @@ let merge_constraint initial_env remove_aliases loc sg constr =
             fun s path -> Subst.add_type_function path ~params ~body s
        in
        let sub = List.fold_left how_to_extend_subst Subst.identity !real_ids in
-       (* This signature will not be used direcly, it will always be freshened
+       (* This signature will not be used directly, it will always be freshened
           by the caller. So what we do with the scope doesn't really matter. But
           making it local makes it unlikely that we will ever use the result of
           this function unfreshened without issue. *)