In semi-arid environments, crops are frequently subjected to a combination of high air temperatures, large atmospheric saturation vapor pressure deficits, high soil temperatures and reduced soil water status. To explore the performance of pearl millet ( Pennisetum typhoides S. and H., cv. CIVT) from panicle initiation to flowering (GS 2) when grown in the field under combinations of these conditions, experiments were conducted in northern Nigeria in three seasons in which daily mean air temperatures during 18 days of this stage averaged 22, 27 and 33°C, and saturation vapor pressure deficits averaged 3.7, 4.0 and 5.2 kPa, respectively. In each experiment, half of the crop was irrigated, while the other half received no water after panicle initiation. For irrigated millet, radiation use efficiency (RUE) did not vary significantly ( P = 0.05) for the three experiments (1.7 g MJ −1). RUE of non-irrigated millet was significantly reduced (0.8 g MJ −1) only during the season with the highest temperature. Radiation interception as a function of thermal time was similar in the irrigated and non-irrigated treatments except in the season with the highest temperatures, when radiation interception was reduced about 25% in the non-irrigated relative to the irrigated treatment. Stem extension of non-irrigated millet did not decline relative to irrigated millet, despite the almost complete extraction of plant available water in the upper 30 cm of the soil, except during the season with the highest temperatures, when stem extension rates began to decline as soon as water was withheld. Under high air temperatures and saturation vapor pressure deficits, dry matter accumulation in both irrigated and non-irrigated millet during GS 2 could be reasonably predicted from RUE and radiation interception. However, when high soil temperatures (daily mean at 5 cm of 34°C) occurred in the non-irrigated treatment, both RUE and radiation interception decreased relative to all other treatments.