diff --git a/math/lcm/lcm.go b/math/lcm/lcm.go
index e1de23e17..1f25728e6 100644
--- a/math/lcm/lcm.go
+++ b/math/lcm/lcm.go
@@ -1,12 +1,33 @@
 package lcm
 
 import (
-	"math"
-
 	"github.com/TheAlgorithms/Go/math/gcd"
 )
 
+// absInt64 returns the absolute value of x as an int64.
+func absInt64(x int64) int64 {
+	if x < 0 {
+		return -x
+	}
+	return x
+}
+
 // Lcm returns the lcm of two numbers using the fact that lcm(a,b) * gcd(a,b) = | a * b |
+//
+// The computation is performed with integer arithmetic and the division by gcd
+// is applied to |a| before multiplying by |b|, so the running intermediate never
+// exceeds the true LCM. This avoids two failure modes of the previous
+// float64-based formula:
+//   - int64 multiply overflow: a*b wraps mod 2^64 before the float64 cast, so
+//     e.g. Lcm(4294967296, 8589934592) returned 0 (impossible for positives).
+//   - float64 precision loss: float64 cannot represent every int64 past 2^53,
+//     so e.g. Lcm(9007199254740993, 1) returned 9007199254740992, violating the
+//     identity lcm(x, 1) == x.
 func Lcm(a, b int64) int64 {
-	return int64(math.Abs(float64(a*b)) / float64(gcd.Iterative(a, b)))
+	g := gcd.Iterative(a, b)
+	if g == 0 {
+		// gcd.Iterative(0, 0) == 0; LCM(0, x) is conventionally 0.
+		return 0
+	}
+	return (absInt64(a) / absInt64(g)) * absInt64(b)
 }
diff --git a/math/lcm/lcm_test.go b/math/lcm/lcm_test.go
index e685a6728..4f7669ae0 100644
--- a/math/lcm/lcm_test.go
+++ b/math/lcm/lcm_test.go
@@ -30,6 +30,35 @@ func TestLcm(t *testing.T) {
 			b:      5,
 			output: 5,
 		},
+		// Regression: integer-arithmetic fix for float precision loss and overflow.
+		{
+			// Mode A: |a*b| overflows int64 even though the true LCM fits.
+			// Previously returned 0 (impossible for positive inputs).
+			name:   "LCM of 2^32 & 2^33 (overflow-safe intermediate)",
+			a:      4294967296,
+			b:      8589934592,
+			output: 8589934592,
+		}, {
+			// Mode A: lcm(x, x) == x. Previously lcm(x,x) != x for large x.
+			name:   "LCM of 3037000500 & 3037000500 (lcm(x,x)==x)",
+			a:      3037000500,
+			b:      3037000500,
+			output: 3037000500,
+		}, {
+			// Mode B: float64 cannot represent 2^53+1 exactly.
+			// Previously returned 9007199254740992, violating lcm(x,1)==x.
+			name:   "LCM of 9007199254740993 & 1 (2^53+1 precision)",
+			a:      9007199254740993,
+			b:      1,
+			output: 9007199254740993,
+		}, {
+			// Mode B: product exceeds 2^53 but fits int64; float64 loses low bits.
+			// Previously returned 1000000016000000000 (off by 63).
+			name:   "LCM of 1000000007 & 1000000009 (large coprime precision)",
+			a:      1000000007,
+			b:      1000000009,
+			output: 1000000016000000063,
+		},
 	}
 
 	for _, tc := range testCases {