We assessed the reliability of anorectal angle (ARA) measurement as an index of fecal incontinence. The "posterior" ARA was measured at rest, squeezing, and straining in 69 continent and 82 incontinent subjects all complaining of various evacuation dysfunctions. The two groups were homogeneous with regard to sex distribution (48.6% vs. 51.4% men and 44.7% vs. 55.3% women, n.s.) and age (56.5 +/- 10.2 vs. 59.3 +/- 9.7 years, n.s.). The incidence of rectal prolapse was the same in the two groups (40 each). The intraobserver agreement index from two independent measurements (Pearson's correlation coefficient), age, and gender interaction [T2 Hotelling test in multivariate analysis of variance (ANOVA)] and the most discriminating category of ARA measurement (Fisher's F test in ANOVA) were calculated. In addition, the relationship between ARA and severity of incontinence was assessed by the eta coefficient. Pearson's correlation coefficient was between 0.78 and 0.98 (P < 0.01). The mean ARA differed significantly between the continent and incontinent subjects (104.5 +/- 10.3 degrees vs. 116.2 +/- 23.6 degrees at rest, 84.5 +/- 14.2 degrees vs. 95.1 +/- 20.1 degrees on squeezing, and 133.7 +/- 21.7 degrees vs. 141.7 +/- 25.9 degrees on straining; T2 0.066, P < 0.05 in multivariate ANOVA). No interaction was noted between groups and gender (T2 = 0.023; F = 1.11, n.s.). Resting ARA was shown by ANOVA to be the most discriminating index (F = 9.4 P < 0.01) between the two groups. Overall, ARA measurement was correlated with the severity of fecal incontinence (eta coefficient: 0.894 at rest; 0.811 on squeezing; 0.695 on straining); its accuracy was 79%, the false-positive rate was 15.3% and the false-negative rate 26.5%. Irrespective of the underlying abnormality, namely rectal prolapse, ARA measurement by defecography can: (a) be reinterpreted reliably by the same observer and (b) differentiate continent from incontinent subjects.