Traffic equilibrium in congested networks is dealt by considering queues at intersections. The travel time on a link with congested flow is expressed as the sum of running time at non-congested flow regime and imaginary waiting time at the end of link. Not only traffic volume but also waiting time in queues are required through equilibrium conditions. The problem is formed as a convex programming, by adding capacity restraints explicitly to the traditional equilibrium problem. The Lagrange multiplier associated with a capacity restraint implies the waiting time in a queue. The problem can be solved numerically by constrained optimization methods. A solution using Lagrange multiplier method and some examples are showed. If queues do not extend to their upstream links, then the solution is unique. Elsewhere, signal control conditions should be added to the problem to require a rational solution.