We examined prospectively risk factors which might contribute to INH-induced liver damage in 113 patients taking preventive INH for at least 8 weeks. Twelve who had abnormal initial liver tests did not get worse with INH, while 19/101 with normal initial tests developed significant liver dysfunction, mostly hepatocellular, three having overt hepatitis. When 12 other patients who drank alcohol were excluded from analysis, there were still 15/89 with significant liver dysfunction, 12 of whom were slow acetylators (p less than 0.05). The only other risk factor was age. By combining acetylator phenotype with age, but excluding alcohol, we calculated the risk of INH-induced liver enzyme elevation as follows: under 35 years--fast acetylators, 3.7%, slow acetylators, 13%; over 35--fast acetylators, 13.2%, slow acetylators, 37% (p less than 0.02). Fast acetylation is thus not a risk factor for developing INH-induced liver dysfunction; indeed, the contrary seems to be the case.