Objective: To compare the effects of food and antacids on the bioavailability of first-line anti-tuberculosis drugs. Method: Systematic search of electronic databases PubMed (January 1950-May 2009), and the Cochrane Library database (January 1974-May 2009), including the Cochrane Centre register of controlled trials, and ongoing trials from research registers using key terms 'food', 'antacids', 'meal', 'controlled trial', 'diet', and the first-line anti-tuberculosis drugs isoniazid (INH), rifampicin (RMP), ethambutol (EMB) and pyrazinamide (PZA). Meta-analysis was performed using RevMan software 5 to assess the impact of food or antacids on the maximum plasma concentrations (C(max)) and area under the plasma concentration time curve (AUC) of anti-tuberculosis drugs. Results: Twelve trials involving 157 patients were included in the meta-analysis. The overall effects showed that food significantly reduced the C(max) mean difference (C(max) MD; C(max) MD -1.42, 95%CI -1.56--1.28, P < 0.00001) and AUC (C(max) MD -3.33, 95%CI -4.05--2.62, P < 0.00001) of INH but antacids did not. Food also significantly reduced the C(max) MD (C(max) MD -2.47, 95%CI -3.30--1.64, P < 0.00001) but not the AUC of RMP. Antacids had no effect on the C(max) MD or AUC of RMP. The C(max) and AUC of PZA were unaffected by both food and antacids. Both food and antacids reduced the C(max) but not the AUC of EMB. Conclusion: From a pharmacokinetic point of view, it seems that the better option for patients with gastrointestinal upsets during chemotherapy would be to add antacids rather than dosing with meals.