diff --git a/lib/std/atomic.c3 b/lib/std/atomic.c3
index 947359829..3a881adf4 100644
--- a/lib/std/atomic.c3
+++ b/lib/std/atomic.c3
@@ -272,7 +272,7 @@ macro fetch_mul(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
 
  @require $defined(*ptr) : "Expected a pointer"
  @require @is_native_atomic_value(*ptr) : "Only types that are native atomic may be used."
- @require $defined(*ptr * y) : "/ must be defined between the values."
+ @require $defined(*ptr / y) : "/ must be defined between the values."
  @require $ordering != NOT_ATOMIC && $ordering != UNORDERED : "Acquire ordering is not valid."
 *>
 macro fetch_div(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
diff --git a/lib/std/core/string_to_real.c3 b/lib/std/core/string_to_real.c3
index 73d27372b..d39e6a03a 100644
--- a/lib/std/core/string_to_real.c3
+++ b/lib/std/core/string_to_real.c3
@@ -158,7 +158,7 @@ macro double? decfloat(char[] chars, int $bits, int $emin, int sign)
 	if (rp % 9)
 	{
 		long rpm9 = rp >= 0 ? rp % 9 : rp % 9 + 9;
-		int p10 = P10S[8 - rpm9];
+		uint p10 = P10S[8 - rpm9];
 		uint carry = 0;
 		for (k = a; k != z; k++)
 		{
diff --git a/lib/std/io/formatter_private.c3 b/lib/std/io/formatter_private.c3
index 164931b0e..1bf83d993 100644
--- a/lib/std/io/formatter_private.c3
+++ b/lib/std/io/formatter_private.c3
@@ -377,13 +377,13 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 		j %= 9;
 		int i;
 		for (i = 10, j++; j < 9; i *= 10, j++);
-		x = *d % i;
+		x = *d % (uint)i;
 		// Are there any significant digits past j?
 		if (x || (d + 1) != z)
 		{
 			double round = 2 / math::DOUBLE_EPSILON;
 			double small;
-			if (((*d / i) & 1) || (i == 1000000000 && d > a && (d[-1] & 1)))
+			if (((*d / (uint)i) & 1) || (i == 1000000000 && d > a && (d[-1] & 1)))
 			{
 				round += 2;
 			}
@@ -437,7 +437,7 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 			// Count trailing zeros in last place
 			if (z > a && z[-1])
 			{
-				for (int i = 10, j = 0; z[-1] % i == 0; i *= 10, j++);
+				for (int i = 10, j = 0; z[-1] % (uint)i == 0; i *= 10, j++);
 			}
 			else
 			{
diff --git a/releasenotes.md b/releasenotes.md
index fff471612..3a8791e15 100644
--- a/releasenotes.md
+++ b/releasenotes.md
@@ -8,6 +8,7 @@
 - Improve support for Android with Termux.
 - Integrated download of the MSVC SDK when compiling for Windows.
 - For `c3c init` with library templates, provide example exported functions. #2898
+- `unsigned % signed` and `unsigned / signed` is no longer allowed without explicit casts, except for const denominators. #2928
 
 ### Stdlib changes
 - Summarize sort macros as generic function wrappers to reduce the amount of generated code. #2831
diff --git a/src/compiler/sema_expr.c b/src/compiler/sema_expr.c
index bd9a352e7..4159ae97c 100644
--- a/src/compiler/sema_expr.c
+++ b/src/compiler/sema_expr.c
@@ -1524,6 +1524,7 @@ static inline bool sema_binary_analyse_arithmetic_subexpr(SemaContext *context,
 		if (flat_left->type_kind == TYPE_BITSTRUCT && left_type == right_type) return true;
 		if (flat_left == type_bool && left_type == right_type) return true;
 	}
+
 	// 2. Perform promotion to a common type.
 	return sema_binary_arithmetic_promotion(context, left, right, left_type, right_type, expr, error,
 		bool_and_bitstruct_is_allowed, operator_overload_ref, failed_ref);
@@ -8157,6 +8158,16 @@ static bool sema_expr_analyse_mult(SemaContext *context, Expr *expr, Expr *left,
 	return true;
 }
 
+static inline bool sema_convert_denominator_to_unsigned_if_needed(SemaContext *context, Expr *left, Expr *right)
+{
+	if (!type_is_unsigned(left->type->canonical) || !type_is_signed(right->type->canonical)) return true;
+	if (sema_cast_const(right) && expr_is_const_int(right) && !int_is_neg(right->const_expr.ixx))
+	{
+		return cast_implicit(context, right, type_int_unsigned_by_bitsize(type_bit_size(right->type->canonical)), true);
+	}
+	return false;
+}
+
 /**
  * Analyse a / b
  * @return true if analysis completed ok.
@@ -8165,6 +8176,11 @@ static bool sema_expr_analyse_div(SemaContext *context, Expr *expr, Expr *left,
 {
 	// 1. Analyse sub expressions and promote to a common type
 	OperatorOverload overload = OVERLOAD_DIVIDE;
+	if (!sema_convert_denominator_to_unsigned_if_needed(context, left, right))
+	{
+		if (failed_ref) return *failed_ref = true, false;
+		RETURN_SEMA_ERROR(expr, "Cannot implicitly divide an unsigned integer by an non-const signed integer, please use explicit casts.");
+	}
 	if (!sema_binary_analyse_arithmetic_subexpr(context, expr, "Cannot divide %s by %s.", false, &overload, failed_ref)) return false;
 	if (!overload) return true;
 
@@ -8221,6 +8237,12 @@ static bool sema_expr_analyse_mod(SemaContext *context, Expr *expr, Expr *left,
 {
 	// 1. Analyse both sides and promote to a common type
 	OperatorOverload overload = OVERLOAD_REMINDER;
+	if (!sema_convert_denominator_to_unsigned_if_needed(context, left, right))
+	{
+		if (failed_ref) return *failed_ref = true, false;
+		RETURN_SEMA_ERROR(expr, "Cannot implicitly convert the values in a remainder operation with an unsigned nominator and non-const or negative signed denominator, please use explicit casts.");
+	}
+
 	if (!sema_binary_analyse_arithmetic_subexpr(context, expr, "Cannot calculate the remainder %s %% %s",
 		false, &overload, failed_ref)) return false;
 	if (!overload) return true;
diff --git a/test/unit/stdlib/atomic.c3 b/test/unit/stdlib/atomic.c3
index b7ba45bb1..942e50db6 100644
--- a/test/unit/stdlib/atomic.c3
+++ b/test/unit/stdlib/atomic.c3
@@ -87,7 +87,7 @@ fn void div() @test
 	{
 		t.create(fn int(void* arg) {
 			thread::sleep_ms(5);
-			atomic::fetch_div(&a, 8);
+			atomic::fetch_div(&a, 8U);
 				return 0;
 		}, null)!!;
 	}