Thirty overweight patients with clinically characterized and biopsy proven nonalcoholic steatohepatitis (NASH) were enrolled in a 48-week treatment trial with rosiglitazone, a peroxisome proliferator-activator receptor (PPAR)-gamma agonist that enhances insulin sensitivity. Improvement in laboratory liver tests, insulin resistance and liver fat content were documented; blinded biopsy review demonstrated decreases in necroinflammatory activity or grade and in individual components of grade, and changes in the relationship of lobular and portal inflammation as well as in the nature of perisinusoidal fibrosis. The current study identified correlations of histological features of the protocol entry biopsy specimens with contemporaneous laboratory and imaging tests. Significant correlations with histologically assessed steatosis were liver fat, evaluated by computed tomography (P = 0.001); mean HbA1C, a measure of glycemic control (P = 0.004); and QUICKI, a measure of insulin sensitivity (P = 0.05). Histologically determined grades of steatohepatitis (SH) correlated with HbA1C (P = 0.01), and a trend toward elevated fasting glucose levels was seen. No subject in the study was cirrhotic at entry; fibrosis scores of the 30 subjects did not significantly correlate with age, gender, body mass index, or clinical tests. All subjects underwent 3 biopsies (prior, entry, and posttreatment), and all had undergone a prior biopsy with diagnostic SH. By blinded analysis, 7 study entry biopsy specimens did not fulfill published strict criteria for SH. Laboratory results from these subjects included normal fasting glucose level and, compared with the 23 subjects with criteria for SH, lower mean alanine aminotransferase and aspartate aminotransferase levels (P = 0.02 for both), less insulin resistance (P = 0.03), and lower mean HbA1C (P = 0.001). We conclude that biopsy findings determined by blinded analysis correlated with image-detected steatosis, laboratory markers of hepatic inflammation, insulin resistance, and long-term glycemia; the findings confirm the usefulness of strict histological criteria in the evaluation of NASH.