Current attempts to find a vaccine for tuberculosis (TB) are based on the assumption that it must drive a Th1 response. We review the evidence that progressive disease might not be due to absence of Th1, but rather to the subversive effect of an unusual Th2-like response, involving interleukin-4 (IL-4) and IL-4delta2. This Th2-like response can impair bactericidal function and lead to toxicity of tumour necrosis factor-alpha (TNF-alpha) and to pulmonary fibrosis. If this is important, effective vaccines will need to suppress pre-existing Th2-like activity. Such vaccines are feasible and are active therapeutically in mouse TB.