Abstract The apparent in vivo substrate specificities for six bacilli desaturases, previously shown to insert a cis-double bond into position five of palmitate and of one which desaturates in position 10 were determined in whole cells by a variety of methods. In addition, the positional specificity of double bond insertion and the rate of temperature-mediated inactivation of these desaturases were also determined. The results can be summarized as follows. In five bacilli strains only one desaturase (Δ5) was found while in one organism (Bacillus licheniformis 9259), both Δ5 and Δ10 desaturation were observed and it was shown that at least two different desaturases were involved. For all desaturases, with the possible exception of the Δ5-desaturase of B. licheniformis 9259, maximal desaturation activity was found with palmitate (n-C16:0) with activity for the n-saturated substrates decreasing in the order C16 g C17 g C18. The Δ10-desaturase of B. licheniformis 9259 showed almost absolute specificity for palmitate while the Δ5-desaturases of the other species showed quantitative variations in their preferences for these three substrates ranging from that of Bacillus megaterium 14581, whose relative desaturation activity was 1.00, 0.57, and 0.16 toward n-C16:0, n-C17:0, and n-C18:0, respectively, to that of Bacillus pumilis 7061 (1.00, 0.81, and 0.74 for the same three fatty acids). Pentadecanoic acid (n-C15:0), tested only in the latter two organisms, had a specificity relative to palmitate of about 0.5 in both cases. The Δ5-desaturase of B. licheniformis 9259 favored n-C17:0 over n-C18:0 but, because of competition with the Δ10-desaturase, its relative activity toward n-C16:0 could not be determined. In the three organisms tested (B. megaterium 14581, B. pumilis 7061, and B. licheniformis 9259) branched chain fatty acids and Δ9 or Δ10 mono-unsaturated fatty acids were also desaturated. 14-Methyl-pentadecanoate (iso-C16:0) was almost as good a substrate as n-C16:0 while iso- and anteiso-C17:0 were less active. However, the iso- and anteiso-C15:0 homologs (the major fatty acids of these organisms) were not desaturated at all. For the three unsaturated substrates tested, the order of activity for Δ5 desaturation was Δ10-C16:1 g g Δ9-C18:1 g Δ9-C16:1. The positional specificity of desaturation, in all bacilli strains, was always the same relative to the carboxyl carbon of the substrate, regardless of substrate chain length, branching, or the presence of a double bond in positions 9 or 10. From these results, we concluded that the substrate is attached to the desaturase at the carboxyl carbon of the fatty acid and that the carboxyl-binding site of the enzyme must be at a fixed distance from the active (hydrogen removal) site to account for the absolute positional specificity of double bond insertion. On the other hand, the efficiency of desaturation (substrate specificity) must depend on a second binding site which anchors the substrate to the enzyme near the terminal methyl group of the fatty acid. This site is visualized as a hydrophobic pocket on the enzyme surface which may vary slightly in shape from enzyme to enzyme but which appears to be located about 15 carbon atoms away from the carboxyl-binding site (assuming that the fatty acid substrate is fully extended along the enzyme surface). Finally, there appears to be no clearly discernible relationship among three desaturase characteristics which presumably depend upon desaturase structure, namely, substrate specificity, positional specificity of double bond insertion, and temperature-mediated first order inactivation of the desaturase. We conclude that these three characteristics may vary independently of each other and that none of the three are directly related to the mechanism of O2-dependent desaturation.