Existing 14, 15 and 16-membered macrolide antibiotics, while effective for other bacterial infections, including some mycobacteria, have not demonstrated significant efficacy in tuberculosis. Therefore an attempt was made to optimize this class for activity against Mycobacterium tuberculosis through semisyntheses and bioassay. Approximately 300 macrolides were synthesized and screened for anti-TB activity. Structural modifications on erythromycin were carried out at positions 3, 6, 9, 11, and 12 of the 14-membered lactone ring; as well as at position 4'' of cladinose and position 2' of desosamine. In general, the synthesized macrolides belong to four subclasses: 9-oxime, 11,12-carbamate, 11,12-carbazate, and 6-O-substituted derivatives. Selected compounds were assessed for mammalian cell toxicity and in some cases were further assessed for CYP3A4 inhibition, microsome stability, in vivo tolerance and efficacy. The activity of 11,12-carbamates and carbazates as well as 9-oximes is highly influenced by the nature of the substitution at these positions. For hydrophilic macrolides, lipophilic substitution may result in enhanced potency, presumably by enhanced passive permeation through the cell envelope. This strategy, however, has limitations. Removal of the C-3 cladinose generally reduces the activity. Acetylation at C-2' or 4'' maintains potency of C-9 oximes but dramatically decreases that of 11,12-substituted compounds. Further significant increases in the potency of macrolides for M. tuberculosis may require a strategy for the concurrent reduction of ribosome methylation.