diff --git a/src/uucore/src/lib/features/format/num_format.rs b/src/uucore/src/lib/features/format/num_format.rs
index e7f9314df65..b1f4095517e 100644
--- a/src/uucore/src/lib/features/format/num_format.rs
+++ b/src/uucore/src/lib/features/format/num_format.rs
@@ -7,12 +7,13 @@
 
 //! Utilities for formatting numbers in various formats
 
-use bigdecimal::BigDecimal;
 use bigdecimal::num_bigint::ToBigInt;
+use bigdecimal::{BigDecimal, RoundingMode};
 use num_traits::Signed;
 use num_traits::Zero;
 use std::cmp::min;
 use std::io::Write;
+use std::num::NonZeroU64;
 
 use super::{
     ExtendedBigDecimal, FormatError, check_precision,
@@ -486,16 +487,18 @@ fn format_float_decimal(
 ///   `.` (the caller must add that to get a valid scientific format number).
 /// - `e` is an integer exponent.
 fn bd_to_string_exp_with_prec(bd: &BigDecimal, precision: usize) -> (String, i64) {
-    // TODO: A lot of time is spent in `with_prec` computing the exact number
+    // TODO: A lot of time is spent in `with_precision_round` computing the exact number
     // of digits, it might be possible to save computation time by doing a rough
     // division followed by arithmetics on `digits` to round if necessary (using
     // `fast_inc`).
 
-    // Round bd to precision digits (including the leading digit)
-    // Note that `with_prec` will produce an extra digit if rounding overflows
-    // (e.g. 9995.with_prec(3) => 1000 * 10^1, but we want 100 * 10^2), we compensate
+    // Round bd to precision digits (including the leading digit), rounding
+    // ties to even like C printf and like `%f` does.
+    // Note that rounding will produce an extra digit if it overflows
+    // (e.g. 9995 rounded to 3 digits => 1000 * 10^1, but we want 100 * 10^2), we compensate
     // for that later.
-    let bd_round = bd.with_prec(precision as u64);
+    let precision_nz = NonZeroU64::new(precision as u64).unwrap_or(NonZeroU64::MIN);
+    let bd_round = bd.with_precision_round(precision_nz, RoundingMode::HalfEven);
 
     // Convert to the form XXX * 10^-p (XXX is precision digit long)
     let (frac, mut p) = bd_round.as_bigint_and_exponent();
@@ -1149,6 +1152,38 @@ mod test {
         assert_eq!(f(99_999_999.0), "1.e+08");
     }
 
+    #[test]
+    fn scientific_and_shortest_float_round_half_to_even() {
+        use super::{format_float_scientific, format_float_shortest};
+        let f = |x, precision| {
+            format_float_scientific(
+                &BigDecimal::from_str(x).unwrap(),
+                Some(precision),
+                Case::Lowercase,
+                ForceDecimal::No,
+            )
+        };
+        assert_eq!(f("2.5", 0), "2e+00");
+        assert_eq!(f("3.5", 0), "4e+00");
+        assert_eq!(f("0.125", 1), "1.2e-01");
+        assert_eq!(f("1.125", 2), "1.12e+00");
+        assert_eq!(f("1.1250001", 2), "1.13e+00");
+        assert_eq!(f("9.5", 0), "1e+01");
+
+        let f = |x, precision| {
+            format_float_shortest(
+                &BigDecimal::from_str(x).unwrap(),
+                Some(precision),
+                Case::Lowercase,
+                ForceDecimal::No,
+            )
+        };
+        assert_eq!(f("0.25", 1), "0.2");
+        assert_eq!(f("0.75", 1), "0.8");
+        assert_eq!(f("2.5", 0), "2");
+        assert_eq!(f("125", 2), "1.2e+02");
+    }
+
     #[test]
     fn hexadecimal_float() {
         // It's important to create the BigDecimal from a string: going through a f64
diff --git a/tests/by-util/test_printf.rs b/tests/by-util/test_printf.rs
index f598b6ab287..0198da2bc0e 100644
--- a/tests/by-util/test_printf.rs
+++ b/tests/by-util/test_printf.rs
@@ -43,6 +43,8 @@ use uutests::new_ucmd;
 #[case::sub_num_sci_trunc(&["pi is ~ %e", "3.1415926535"], "pi is ~ 3.141593e+00")]
 #[case::sub_num_dec_trunc(&["pi is ~ %g", "3.1415926535"], "pi is ~ 3.14159")]
 #[case::sub_num_sci_negative(&["-1234 is %e", "-1234"], "-1234 is -1.234000e+03")]
+#[case::sub_num_sci_round_half_even(&["%.1e %.0e %.0E %.2e %.0e", "0.125", "2.5", "-2.5", "1.125", "3.5"], "1.2e-01 2e+00 -2E+00 1.12e+00 4e+00")]
+#[case::sub_num_general_round_half_even(&["%.1g %.0g %.2g", "0.25", "2.5", "125"], "0.2 2 1.2e+02")]
 #[case::sub_num_hex_float_lower(&["%a", ".875"], "0xep-4")]
 #[case::sub_num_hex_float_upper(&["%A", ".875"], "0XEP-4")]
 #[case::sub_min_width(&["hello %7s", "world"], "hello   world")]
diff --git a/tests/by-util/test_seq.rs b/tests/by-util/test_seq.rs
index 8a081709942..115e02d3264 100644
--- a/tests/by-util/test_seq.rs
+++ b/tests/by-util/test_seq.rs
@@ -859,6 +859,14 @@ fn test_format_option_default_precision_scientific() {
         .stdout_only("0.000000E+00\n7.000000E-01\n1.400000E+00\n");
 }
 
+#[test]
+fn test_format_option_scientific_round_half_even() {
+    new_ucmd!()
+        .args(&["-f", "%.0e", "2.5", "1", "3.5"])
+        .succeeds()
+        .stdout_only("2e+00\n4e+00\n");
+}
+
 #[test]
 fn test_auto_precision() {
     new_ucmd!()