We propose a novel stochastic radio resource allocation strategy for the uplink that achieves long-term fairness in terms of similar bitrates considering backhaul and air-interface capacity limitations. We focus on a single cell scenario based on WCDMA technology. We propose to use a maximin criteria to introduce fairness among the different users' throughputs. An stochastic approximation is implemented to obtain an online algorithm where the Lagrange multipliers are estimated at each scheduling period. Our results show that the proposed scheme achieves higher fairness among the users and, in some cases, a higher sum-rate compared with the well-known proportional fair scheduler.