Potassium (K) fertilisation at planting time may not be sufficient to achieve high cassava yields. Experiments were conducted over two growing periods using sweet cassava (SC) variety IAC 576-70 and bitter cassava (BC) variety IAC 13 aiming to evaluate the effects of K fertilisation rate (0-150 kg K/ha) and timing of application (one, two, or three) on leaf K concentration, yield components, and storage-root and starch yields. The SC and BC varieties were grown in a randomized block design in a 3 × 4 + 1 factorial scheme with four replications during growing periods of 10 and 20 months respectively. For SC, split application of K at planting plus 3 months after planting (MP) or at planting plus 1.5 and 3 MP was more effective at increasing the marketable root yield, with greater rainfall at the beginning of the growing period. Split application of K at planting plus 3 MP also reduced the cooking time and cooked root firmness. For BC, split application of K at planting plus 12 MP more effectively increased the storage-root and starch yields; however, root yield response to K fertilisation in the second growing period, with greater rain intensity during the initial crop stages, was greater than in the first growing period. In K-deficient soils, the optimal K application rate for SC marketable root yield varied from 79 to 111 kg/ha when K was applied after 1.5 MP, and linear responses to K application rate occurred when K was applied earlier. For BC, the optimal K application rate varied from 75 to 92 kg K/ha. Our data suggest that K application times should be different for cassava varieties with short and long growing periods.