In a multi-input multi-output (MIMO) interference network, the maximum achievable signal-to-interference-plusnoise ratio (SINR) determines the level of quality of service (QoS) that can be supported by each stream. In this paper, we study the problem of maximizing the minimum SINR of independent parallel data streams across all users in a MIMO interference network that employs interference alignment (IA) in the presence of channel state information (CSI) uncertainty. To this end, we propose a low-complexity robust SINR fairness algorithm based on the power allocation across data streams. Building on the conventional IA design, we provide a closed-form solution for the power allocation problem. It is observed that the proposed power allocation (PA)-based robust design outperforms the conventional robust design , when the channel changes drastically and the CSI uncertainty cannot be reliably bounded. Furthermore, the PAbased robust design provides significant performance gain over the benchmark Max-SINR algorithm.