The optimal combination of hemodialysis (HD) dose and session length remains uncertain, and previous studies have not conclusively shown session length to be an important independent determinant of patient mortality. The objective of this study was to examine associations between HD dose and session length with mortality risk using data from the Australian and New Zealand Dialysis and Transplant Registry. Analyses were performed using a prospective inception cohort comprising all incident adult patients treated by thrice-weekly maintenance HD, who commenced renal replacement therapy with HD between 1 April 1997 and 31 March 2004. In all, 6593 patients were identified, of whom 4193 had sufficient data for multivariate analyses. HD dose (single pool fractional clearance of urea, Kt/V) and session length were included in analyses as those recorded 12 months after HD inception to reduce confounding by residual renal function. The outcome examined was patient mortality. Survival analyses included Kaplan-Meier calculations of survival and Cox regression for multivariate analyses. Covariates in Cox models included patient demographics, co-morbid medical conditions at HD inception, and HD operating parameters. After adjustment for covariates and each other, Kt/V of 1.30-1.39 and session length of 4.5-4.9 h were associated with the lowest mortality risk. There was no interaction between HD dose and session length. Thus, the optimal combination for mortality appears to be Kt/V of > or = 1.3 and session length of > or = 4.5 h. These data suggest a randomized controlled trial to test these hypotheses, and support the inclusion of criteria relating to session length in definitions of adequate HD practice.