Early studies of peak-period traffic congestion generally ignored the departure time decision of road users. Vickrey (1969) and Kocur (1981) remedied this oversight by deriving the departure rate along a single route subject to queuing congestion as the outcome of individual utility maximization. Arnott et al (1985) extend their work, examining alternative toll schemes and deriving rules for optimal capacity. This paper extends these analyses to a simple network of routes in parallel.