Two side-by-side energy windows are sometimes employed in quantitative single photon emission computed tomography (SPECT) studies. The count-rate losses at high activities for a GE400AT camera were measured in such a dual-window-acquisition mode by imaging a decaying source composed of a hot sphere within a warm cylinder. The data in each window was either kept separate or combined for purposes of dead-time correction and the paralyzable model was assumed. In addition, correction factors were derived from a "monitor" source at the edge of the camera. Finally, energy spectra for only the monitor-source region of interest and for the entire camera were obtained under both low- and high-count conditions and compared. With 99mTc (1) the spectral measurements show no peak shift but reveal pulse-pile-up spectral degradation. (2) The monitor-source corrections do not agree well with those from the model, presumably because of the differential effects of such degradation. For this camera the preferred correction method for patients is one using the model and two, effective, phantom-derived dead times for the separate data from the two windows. The two effective dead times are needed to compensate for pulse pile-up adding more counts to the lower-energy window than to the higher-energy one at high rates.(ABSTRACT TRUNCATED AT 250 WORDS)