Satellite observations of outgoing terrestrial infrared (IR) radiation as a function of latitude exhibit a minimum near the equator 20-40 Wm-' smaller than peaks in the subtropics. We attempt to dissect the causes of the dip through calculations with a spectrallydetailed multi-level radiative transfer model. Roughly one third of the dip can be attributed to the latitudinal variation of atmospheric water vapour; the remainder apparently is due to latitudinal variations in cloud amount and (especially) cloud-top height. However, inclusion of clouds as given by published climatologies enhances the clear-sky dip only slightly. Thus, about one half of the dip is essentially unexplained. We suspect the explanation is that near-equatorial cirrus and cumulonimbus heights are too low in the cloud climatology, emphasizing the need for a better cloud climatology.Since tropospheric humidity variations have a strong effect on clear-sky outgoing IR, empirical studies which correlate cloud with IR variations may overestimate the effect of clouds on outgoing IR if cloud amount is correlated with relative humidity.The effect of humidity variations on outgoing IR suggests that a measure of tropospheric humidity be incorporated explicitly in the parametrization of outgoing IR for simple climate models. Sci., 31, 435446. AK, 245-254. J . Quant. Spectr. and Rad. Trans., 16,447-489. Comp. Phys., 24,416-444.