The ability to change direction is crucial in soccer due to the high frequency of rapid movement adjustments during match play. This study examined the effects of change-of-direction (COD) angle (45°, 90°,  180°) and limb dominance on sprint speed in 18 experienced male soccer players using a multi-angle field based assessment. Athletes completed a 20 m linear sprint and three 20 m COD sprints, each incorporating a single directional change at 15 m. Mean speed was assessed across the 0–10 m, 10–20 m (COD speed), 10–13 m (entry speed), and 13–20 m segments, and percentage COD speed deficit relative to linear sprinting (%CODD) was calculated. Mean speed in the 0–10 m segment was identical between linear and COD sprints (5.18 ± 0.28 to 5.28 ± 0.28 m · s−1). Using the linear 0–10 m speed as a comparator, %CODD values were positive or near-zero for 45° and 90° (+21.0 ± 12.7% and −2.9 ± 7.7%, respectively), indicating greater COD speed than linear, with a negative value only for 180° (−25.2 ± 4.3%). When COD ability was quantified relative to the 10–20 m segment of the linear sprint (7.53 ± 0.37 m · s−1), a clear and progressively greater %CODD was observed across angles (45°: −15.6 ± 7.6%; 90°: −32.2 ± 4.7%; 180°: −47.7 ± 2.7%; all p < 0.001). Entry speed declined from 45° to 90° COD (6.55 ± 0.45 to 5.77 ± 0.34 m · s−1, p < 0.001), with no difference between 90° and 180° (5.79 ± 0.31 m · s−1, p = 1.0). Kendall’s tau correlations showed moderate agreement in %CODD between adjacent angles, but not between 45° and 180°, with no effect of limb dominance (all p > 0.05). In conclusion, %CODD should be calculated using matched linear segments. Angle-dependent and individual variability in %CODD support multi-angle, rather than single-direction, assessment and training.