The effects of temperature and daylength on durations from sowing to flowering (f) were studied in six short-duration genotypes of pigeonpea (Cajanus cajan) which had been classified as extra-early or early to mature at ICRISAT Center (17 degrees N). Plants were grown in seven locations in Kenya covering a wide range of altitudes (50-2000 m) at latitudes ranging from 0 degrees to 4 degrees S. Experiments spanned two seasons, and were conducted in normal field conditions as well as under clear polythene enclosures constructed at six of the sites in order to provide warmer than ambient temperatures. All genotypes were also sown at monthly intervals as well as under an artificially extended daylength, both at Katumani (1 degrees 30'S). Mean pre-flowering values of temperature (T) and photoperiod (P) varied from 15.6 degrees to 34.0 degrees C and from 12.6 to 15.0 h, respectively. These photothermal conditions resulted in values of f between 53 and 118 d. Rates of progress from sowing to flowering (1/f) were mostly unaffected by but were sometimes slightly responsive to photoperiod, whereas they responded consistently strongly to mean pre-flowering temperature ( T) below and above an optimum value close to 24 degrees C. In the sub-optimal range the effects of T were positive and in the supra-optimal range they were negative, such that 1/f=a+bT and 1/f=a'-b'T, respectively. The genotype-specific parameters a, b, a' and b' from these linear-rate models based on flowering responses in 27 environments were validated with independent data. Predicted values in f in 12 monthly sowings (range 62-99 d) did not differ significantly from those observed, i.e. for practical purposes within Kenya, the flowering responses of genotypes classified as extra-early or early to mature are insensitive to photoperiod. At more extreme latitudes (e.g. in India), and given the slight photoperiod sensitivity of some early-maturing genotypes, longer days can combine with supra-optimal temperatures to delay flowering.