This phantom study investigates whether attenuation maps from transmission data degraded by increased noise from subtraction of emission counts can still provide useful attenuation correction in the regular and obese chest. Technetium-99m was used for both emission and transmission on a triple head simultaneous emission transmission tomography (Tc-Tc SETT) system. Fanbeam transmission counts were computed by subtracting emission counts estimated from the two parallel collimator heads. Radioactive decay was used to simulate organ counts from injections of 900 and 400 MBq sestamibi for regular and obese chest sizes. Line source activity was 350 MBq. Control attenuation maps were obtained with no emission activity. Noise control included catering for negative and zero transmission counts, pre-filtering and segmentation of mu maps. Pre-filtering was tried before and after subtraction and before and after setting negative pixels to zero. Mean+/-SD count/pixel at the heart in anterior transmission projections was typically 33+/-18 for the regular and 1+/-7 for the obese chest. For the obese chest, pre-filtering before resetting negative counts best preserved mean mu in soft tissue and lung. Tc-Tc SETT mu mean+/-SD for the regular chest were 0.144+/-0.012 and 0.058+/-0.004 for soft tissue and lung and for the obese chest, 0.152+/-0.075 and 0.059+/-0.017. The accuracy of the Tc-Tc SETT bullseye plots for the regular chest was the same as with control map attenuation correction and 3 times better than with no correction. For the obese chest it was as good as with control map correction only if mu map segmentation was applied. Tc-Tc SETT soft tissue and lung mu in 28 patient studies indicated that segmentation is practical for a wide range of chest sizes. Tc-Tc SETT on a triple-head system offers an accurate, inexpensive method of attenuation correction for the majority of chest sizes.