Although frontal lobe interconnections of the primary (area 4 or M1) and supplementary (area 6m or M2) motor cortices are well understood, how frontal granular (or prefrontal) cortex influences these and other motor cortices is not. Using fluorescent dyes in rhesus monkeys, we investigated the distribution of frontal lobe inputs to M1, M2, and the cingulate motor cortex (area 24c or M3, and area 23c). M1 received input from M2, lateral area 6, areas 4C and PrCO, and granular area 12. M2 received input from these same areas as well as M1; granular areas 45, 8, 9, and 46; and the lateral part of the orbitofrontal cortex. Input from the ventral part of lateral area 6, area PrCO, and frontal granular cortex targeted only the ventral portion of M1, and primarily the rostral portion of M2. In contrast, M3 and area 23c received input from M1, M2; lateral area 6 and area 4C; granular areas 8, 12, 9, 46, 10, and 32; as well as orbitofrontal cortex. Only M3 received input from the ventral part of lateral area 6 and areas PrCO, 45, 12vl, and the posterior part of the orbitofrontal cortex. This diversity of frontal lobe inputs, and the heavy component of prefrontal input to the cingulate motor cortex, suggests a hierarchy among the motor cortices studied. M1 receives the least diverse frontal lobe input, and its origin is largely from other agranular motor areas. M2 receives more diverse input, arising primarily from agranular motor and prefrontal association cortices. M3 and area 23c receive both diverse and widespread frontal lobe input, which includes agranular motor, prefrontal association, and frontal limbic cortices. These connectivity patterns suggest that frontal association and frontal limbic areas have direct and preferential access to that part of the corticospinal projection which arises from the cingulate motor cortex.