Substrate reaction forces (SRFs) were quantified for 502 forelimb and hindlimb contacts in two primate species, Eulemur fulvus and Lemur catta. Three individuals of each taxon were encouraged to traverse a force transducer. Reaction forces were evaluated for linear locomotion on runway and simulated arboreal substrates. Preferred gaits differed between substrates: galloping on the runway and walking on simulated arboreal substrates. Unlike vertical and fore–aft (FA) forces, mediolateral (ML) forces during linear locomotion were relatively independent of speed, which suggests balancing the centre of body mass in the coronal plane was not speed related. Since peak ML forces occasionally exceeded peak FA forces, ML forces were not of trivial importance during primate quadrupedal locomotion. Both lemurids regularly experienced lower ML peak forces during ‘arboreal’ than ‘terrestrial’ quadrupedalism, probably to minimize side-to-side fluctuations of their centre of mass over the simulated arboreal branch supports. Eulemur fulvus encountered greater ML peak forces in their forelimbs, while L. catta encountered greater hindlimb peak ML forces. Lemurids most frequently exerted medially directed peak side-to-side forces during arboreal quadrupedalism and laterally directed peak side-to-side forces during terrestrial quadrupedalism. The direction of hindlimb ML peak force, however, was less consistent than forelimb ML peak force regardless of substrate. The lemurs experienced oscillations in the direction of the forelimb and hindlimb ML force over the duration of support phase, suggesting that both pairs of limbs contributed to balance. More frequent changes in ML force direction of the hindlimb support the predominance of these limbs in keeping side-to-side balance. The results of our study suggest that ML forces warrant consideration in locomotion studies, especially when evaluating joint moments and bending regimes of long bones.