The pharmaceutically active compound theophylline (T) was cocrystallized with the amides formamide (1), acetamide (2), N-methylformamide (3), N,N-dimethylformamide (4), benzamide (5), and pyrazinamide (6), with systems T:1, T:5, and T:6 displaying polymorphic behavior. The cocrystals with formamide (T:1), acetamide (T:2), and benzamide (T:5), and one polymorph of the cocrystal with pyrazinamide (T:6-I), contain an R 2 2 (9) hydrogen bonding motif between the amide cocrystal formers and the HN−C−CO moiety of the theophylline molecule (an amide-pseudo amide synthon). This motif was, however, absent from the other polymorph of the pyrazinamide cocrystal (T:6-II) and also from the N-methylformamide cocrystal (T:3) (and is not possible in the N,N-dimethylformamide cocrystal (T:4)). These observations are rationalized using hydrogen bond propensity calculations, although limitations of using such calculations for predicting cocrystallization are noted. The amide-pseudo amide synthon is favored when theophylline cocrystallizes with both primary amides and with secondary amides which are locked in a cis configuration. On heating, all cocrystals were found to dissociate before melting due to loss of the amide, making stability to dissociation a more meaningful measure of cocrystal stability than melting point for these systems. On dissociation of the cocrystals, theophylline typically crystallizes as the commonly observed polymorph Form II. In the case of the acetamide cocrystal (T:2), however, the rarely observed metastable polymorph, Form V, crystallizes concomitantly with Form II suggesting that cocrystal dissociation on heating could be a strategy for identifying novel polymorphic forms of compounds.