Xenon CT (Xe-CT) is used to estimate regional ventilation by measuring regional attenuation changes over multiple breaths while re-breathing a constant [Xe]. Xe-CT has potential human applications, though anesthetic properties limit [Xe] to 35%. We investigate affects of lower [Xe], including a low [Xe]-krypton combination on time constant (TC) determination. Six anesthetized sheep were scanned prone and supine using multidetector-row CT (MDCT). Lungs were imaged by respiratory gating during washin of a 30%, 40%, 55% Xe, and a 30%Xe/30%Kr mixture. Using Kr avoids unwanted effects of Xe. Mean TCs, coefficients of variation (CV), and half confidence intervals (CI)/Mean served as indexes of sensitivity to noise. Mean supine and prone TCs of three [Xe]s were not significantly different. Average CVs of TCs increased from 57% (55% Xe), 58% (40% Xe) and 73% (30% Xe) (P < 0.05: paired t tests; 30% Xe vs higher [Xe]). Monte Carlo simulation indicated a CV based on inherent image noise was 8% for 55% Xe and 17% for 30% Xe (P< 0.05). Adding 30% Kr to 30% Xe gave a washin signal equivalent to 40% Xe. Half CI/Mean using the 30%Xe/30%Kr mixture was not significantly different from 55 and 40% Xe. Though average TCs were not affected by changes in [Xe], the higher CV and half CI/Mean suggested reduced signal/noise at the lower [Xe]. 30%Xe/30%Kr mixture was comparable to that of 40% Xe, providing an important agent for CT-based assessment of regional ventilation in humans.