diff --git a/core/thread/thread.odin b/core/thread/thread.odin
index 9ba03203f..9fcc5b84f 100644
--- a/core/thread/thread.odin
+++ b/core/thread/thread.odin
@@ -116,26 +116,21 @@ run_with_data :: proc(data: rawptr, fn: proc(data: rawptr), init_context: Maybe(
 }
 
 run_with_poly_data :: proc(data: $T, fn: proc(data: T), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal)
-	where size_of(T) <= size_of(rawptr) {
+	where size_of(T) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	create_and_start_with_poly_data(data, fn, init_context, priority, true)
 }
 
 run_with_poly_data2 :: proc(arg1: $T1, arg2: $T2, fn: proc(T1, T2), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal)
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	create_and_start_with_poly_data2(arg1, arg2, fn, init_context, priority, true)
 }
 
 run_with_poly_data3 :: proc(arg1: $T1, arg2: $T2, arg3: $T3, fn: proc(arg1: T1, arg2: T2, arg3: T3), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal)
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr),
-	      size_of(T3) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) + size_of(T3) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	create_and_start_with_poly_data3(arg1, arg2, arg3, fn, init_context, priority, true)
 }
 run_with_poly_data4 :: proc(arg1: $T1, arg2: $T2, arg3: $T3, arg4: $T4, fn: proc(arg1: T1, arg2: T2, arg3: T3, arg4: T4), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal)
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr),
-	      size_of(T3) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) + size_of(T3) + size_of(T4) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	create_and_start_with_poly_data4(arg1, arg2, arg3, arg4, fn, init_context, priority, true)
 }
 
@@ -178,7 +173,7 @@ create_and_start_with_data :: proc(data: rawptr, fn: proc(data: rawptr), init_co
 }
 
 create_and_start_with_poly_data :: proc(data: $T, fn: proc(data: T), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal, self_cleanup := false) -> ^Thread
-	where size_of(T) <= size_of(rawptr) {
+	where size_of(T) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	thread_proc :: proc(t: ^Thread) {
 		fn := cast(proc(T))t.data
 		assert(t.user_index >= 1)
@@ -188,96 +183,118 @@ create_and_start_with_poly_data :: proc(data: $T, fn: proc(data: T), init_contex
 	t := create(thread_proc, priority)
 	t.data = rawptr(fn)
 	t.user_index = 1
+
 	data := data
-	mem.copy(&t.user_args[0], &data, size_of(data))
+
+	mem.copy(&t.user_args[0], &data, size_of(T))
+
 	if self_cleanup {
 		t.flags += {.Self_Cleanup}
 	}
+
 	t.init_context = init_context
 	start(t)
 	return t
 }
 
 create_and_start_with_poly_data2 :: proc(arg1: $T1, arg2: $T2, fn: proc(T1, T2), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal, self_cleanup := false) -> ^Thread
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	thread_proc :: proc(t: ^Thread) {
 		fn := cast(proc(T1, T2))t.data
 		assert(t.user_index >= 2)
-		arg1 := (^T1)(&t.user_args[0])^
-		arg2 := (^T2)(&t.user_args[1])^
+		
+		user_args := mem.slice_to_bytes(t.user_args[:])
+		arg1 := (^T1)(raw_data(user_args))^
+		arg2 := (^T2)(raw_data(user_args[size_of(T1):]))^
+
 		fn(arg1, arg2)
 	}
 	t := create(thread_proc, priority)
 	t.data = rawptr(fn)
 	t.user_index = 2
+
 	arg1, arg2 := arg1, arg2
-	mem.copy(&t.user_args[0], &arg1, size_of(arg1))
-	mem.copy(&t.user_args[1], &arg2, size_of(arg2))
+	user_args := mem.slice_to_bytes(t.user_args[:])
+
+	n := copy(user_args,     mem.ptr_to_bytes(&arg1))
+	_  = copy(user_args[n:], mem.ptr_to_bytes(&arg2))
+
 	if self_cleanup {
 		t.flags += {.Self_Cleanup}
 	}
+
 	t.init_context = init_context
 	start(t)
 	return t
 }
 
 create_and_start_with_poly_data3 :: proc(arg1: $T1, arg2: $T2, arg3: $T3, fn: proc(arg1: T1, arg2: T2, arg3: T3), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal, self_cleanup := false) -> ^Thread
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr),
-	      size_of(T3) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) + size_of(T3) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	thread_proc :: proc(t: ^Thread) {
 		fn := cast(proc(T1, T2, T3))t.data
 		assert(t.user_index >= 3)
-		arg1 := (^T1)(&t.user_args[0])^
-		arg2 := (^T2)(&t.user_args[1])^
-		arg3 := (^T3)(&t.user_args[2])^
+
+		user_args := mem.slice_to_bytes(t.user_args[:])
+		arg1 := (^T1)(raw_data(user_args))^
+		arg2 := (^T2)(raw_data(user_args[size_of(T1):]))^
+		arg3 := (^T3)(raw_data(user_args[size_of(T1) + size_of(T2):]))^
+
 		fn(arg1, arg2, arg3)
 	}
 	t := create(thread_proc, priority)
 	t.data = rawptr(fn)
 	t.user_index = 3
+
 	arg1, arg2, arg3 := arg1, arg2, arg3
-	mem.copy(&t.user_args[0], &arg1, size_of(arg1))
-	mem.copy(&t.user_args[1], &arg2, size_of(arg2))
-	mem.copy(&t.user_args[2], &arg3, size_of(arg3))
+	user_args := mem.slice_to_bytes(t.user_args[:])
+
+	n := copy(user_args,     mem.ptr_to_bytes(&arg1))
+	n += copy(user_args[n:], mem.ptr_to_bytes(&arg2))
+	_  = copy(user_args[n:], mem.ptr_to_bytes(&arg3))
+
 	if self_cleanup {
 		t.flags += {.Self_Cleanup}
 	}
+
 	t.init_context = init_context
 	start(t)
 	return t
 }
 create_and_start_with_poly_data4 :: proc(arg1: $T1, arg2: $T2, arg3: $T3, arg4: $T4, fn: proc(arg1: T1, arg2: T2, arg3: T3, arg4: T4), init_context: Maybe(runtime.Context) = nil, priority := Thread_Priority.Normal, self_cleanup := false) -> ^Thread
-	where size_of(T1) <= size_of(rawptr),
-	      size_of(T2) <= size_of(rawptr),
-	      size_of(T3) <= size_of(rawptr) {
+	where size_of(T1) + size_of(T2) + size_of(T3) + size_of(T4) <= size_of(rawptr) * MAX_USER_ARGUMENTS {
 	thread_proc :: proc(t: ^Thread) {
 		fn := cast(proc(T1, T2, T3, T4))t.data
 		assert(t.user_index >= 4)
-		arg1 := (^T1)(&t.user_args[0])^
-		arg2 := (^T2)(&t.user_args[1])^
-		arg3 := (^T3)(&t.user_args[2])^
-		arg4 := (^T4)(&t.user_args[3])^
+
+		user_args := mem.slice_to_bytes(t.user_args[:])
+		arg1 := (^T1)(raw_data(user_args))^
+		arg2 := (^T2)(raw_data(user_args[size_of(T1):]))^
+		arg3 := (^T3)(raw_data(user_args[size_of(T1) + size_of(T2):]))^
+		arg4 := (^T4)(raw_data(user_args[size_of(T1) + size_of(T2) + size_of(T3):]))^
+
 		fn(arg1, arg2, arg3, arg4)
 	}
 	t := create(thread_proc, priority)
 	t.data = rawptr(fn)
 	t.user_index = 4
+
 	arg1, arg2, arg3, arg4 := arg1, arg2, arg3, arg4
-	mem.copy(&t.user_args[0], &arg1, size_of(arg1))
-	mem.copy(&t.user_args[1], &arg2, size_of(arg2))
-	mem.copy(&t.user_args[2], &arg3, size_of(arg3))
-	mem.copy(&t.user_args[3], &arg4, size_of(arg4))
+	user_args := mem.slice_to_bytes(t.user_args[:])
+
+	n := copy(user_args,     mem.ptr_to_bytes(&arg1))
+	n += copy(user_args[n:], mem.ptr_to_bytes(&arg2))
+	n += copy(user_args[n:], mem.ptr_to_bytes(&arg3))
+	_  = copy(user_args[n:], mem.ptr_to_bytes(&arg4))
+
 	if self_cleanup {
 		t.flags += {.Self_Cleanup}
 	}
+
 	t.init_context = init_context
 	start(t)
 	return t
 }
 
-
 _select_context_for_thread :: proc(init_context: Maybe(runtime.Context)) -> runtime.Context {
 	ctx, ok := init_context.?
 	if !ok {
diff --git a/tests/core/Makefile b/tests/core/Makefile
index 919262f85..3fdadc246 100644
--- a/tests/core/Makefile
+++ b/tests/core/Makefile
@@ -3,7 +3,7 @@ PYTHON=$(shell which python3)
 
 all: download_test_assets image_test compress_test strings_test hash_test crypto_test noise_test encoding_test \
 	 math_test linalg_glsl_math_test filepath_test reflect_test os_exit_test i18n_test match_test c_libc_test net_test \
-	 fmt_test
+	 fmt_test thread_test
 
 download_test_assets:
 	$(PYTHON) download_assets.py
@@ -61,3 +61,6 @@ net_test:
 
 fmt_test:
 	$(ODIN) run fmt -out:test_core_fmt
+
+thread_test:
+	$(ODIN) run thread -out:test_core_thread
diff --git a/tests/core/build.bat b/tests/core/build.bat
index 1d146c8a4..311c9c78e 100644
--- a/tests/core/build.bat
+++ b/tests/core/build.bat
@@ -85,3 +85,8 @@ echo ---
 echo Running core:container tests
 echo ---
 %PATH_TO_ODIN% run container %COMMON% %COLLECTION% -out:test_core_container.exe || exit /b
+
+echo ---
+echo Running core:thread tests
+echo ---
+%PATH_TO_ODIN% run thread %COMMON% %COLLECTION% -out:test_core_thread.exe || exit /b
diff --git a/tests/core/thread/test_core_thread.odin b/tests/core/thread/test_core_thread.odin
new file mode 100644
index 000000000..c0c7396a7
--- /dev/null
+++ b/tests/core/thread/test_core_thread.odin
@@ -0,0 +1,84 @@
+package test_core_thread
+
+import "core:testing"
+import "core:thread"
+import "core:fmt"
+import "core:os"
+
+TEST_count := 0
+TEST_fail  := 0
+
+t := &testing.T{}
+
+when ODIN_TEST {
+    expect  :: testing.expect
+    log     :: testing.log
+} else {
+    expect  :: proc(t: ^testing.T, condition: bool, message: string, loc := #caller_location) {
+        TEST_count += 1
+        if !condition {
+            TEST_fail += 1
+            fmt.printf("[%v] %v\n", loc, message)
+            return
+        }
+    }
+    log     :: proc(t: ^testing.T, v: any, loc := #caller_location) {
+        fmt.printf("[%v] ", loc)
+        fmt.printf("log: %v\n", v)
+    }
+}
+
+main :: proc() {
+	poly_data_test(t)
+
+	if TEST_fail > 0 {
+		os.exit(1)
+	}
+}
+
+@(test)
+poly_data_test :: proc(_t: ^testing.T) {
+	MAX :: size_of(rawptr) * thread.MAX_USER_ARGUMENTS
+
+	@static poly_data_test_t: ^testing.T
+	poly_data_test_t = _t
+
+	b: [MAX]byte = 8
+	t1 := thread.create_and_start_with_poly_data(b, proc(b: [MAX]byte) {
+		b_expect: [MAX]byte = 8
+		expect(poly_data_test_t, b == b_expect, "thread poly data not correct")
+	})
+	defer free(t1)
+
+	b1: [3]uintptr = 1
+	b2: [MAX / 2]byte = 3
+	t2 := thread.create_and_start_with_poly_data2(b1, b2, proc(b: [3]uintptr, b2: [MAX / 2]byte) {
+		b_expect: [3]uintptr = 1
+		b2_expect: [MAX / 2]byte = 3
+		expect(poly_data_test_t, b == b_expect,   "thread poly data not correct")
+		expect(poly_data_test_t, b2 == b2_expect, "thread poly data not correct")
+	})
+	defer free(t2)
+
+	t3 := thread.create_and_start_with_poly_data3(b1, b2, uintptr(333), proc(b: [3]uintptr, b2: [MAX / 2]byte, b3: uintptr) {
+		b_expect: [3]uintptr = 1
+		b2_expect: [MAX / 2]byte = 3
+
+		expect(poly_data_test_t, b == b_expect,   "thread poly data not correct")
+		expect(poly_data_test_t, b2 == b2_expect, "thread poly data not correct")
+		expect(poly_data_test_t, b3 == 333,       "thread poly data not correct")
+	})
+	defer free(t3)
+
+	t4 := thread.create_and_start_with_poly_data4(uintptr(111), b1, uintptr(333), u8(5), proc(n: uintptr, b: [3]uintptr, n2: uintptr, n4: u8) {
+		b_expect: [3]uintptr = 1
+
+		expect(poly_data_test_t, n == 111,        "thread poly data not correct")
+		expect(poly_data_test_t, b == b_expect,   "thread poly data not correct")
+		expect(poly_data_test_t, n2 == 333,       "thread poly data not correct")
+		expect(poly_data_test_t, n4 == 5,         "thread poly data not correct")
+	})
+	defer free(t4)
+
+	thread.join_multiple(t1, t2, t3, t4)
+}