Jenkins et al1 are to be commended on achieving significant reductions in the duration of antimicrobial therapy and decreased use of broad spectrum agents among patients hospitalized for uncomplicated skin infections, through the use of a management algorithm. However, as noted in the accompanying commentary with respect to the authors' treatment algorithm, more emphasis may be warranted on the “uncomplicated” study inclusion criterion.2 In particular, the authors' advocacy for avoiding (presumably unnecessary) microbiological tests should not be applied inappropriately to other populations with skin infections, in whom such tests, particularly wound cultures, can be extremely helpful in selecting step-down oral therapy. For example, group B streptococci, a not uncommon cause of foot infections among patients with diabetes mellitus or peripheral vascular disease, are usually resistant to tetracyclines, including doxycycline. Likewise, group A streptococci are unreliably responsive to trimethoprim-sulfamethoxazole. Similarly, many methicillin-resistant staphylococci are clindamycin resistant, and some (albeit a small minority) are resistant to tetracyclines and/or trimethoprim-sulfamethoxazole. Moreover, diabetic foot infections, particularly more severe episodes, may involve gram-negative bacilli, which have highly unpredictable susceptibility patterns. Consequently, swab cultures from infected skin ulcers (notwithstanding their nonspecificity) and cultures of abscess pus can yield clinically important information in the appropriate context. The favorable emphasis Jenkins et al1 give in the abstract to the observed “significant decrease in use of microbiological cultures” (which refers primarily to blood cultures, not wound cultures) might lead the casual reader to conclude, one hopes erroneously, that these authors believe wound (or pus) cultures should be avoided generally in patients with cellulitis or skin abscesses. On the contrary, in this era of increasing antimicrobial resistance, wider use of such cultures probably should be encouraged, particularly with compromised hosts, to allow just the sort of targeted therapy Jenkins et al1 appropriately support.