To realize the next generation of ultra-reliable low-latency communications (URLLC) with less radio resource consumption, methods relying solely on the addition of redundancy do not suffice. An alternative is to monitor the wireless channel and prevent outages due to small scale fading by allocating users to suitable radio resources based on the channel gain. To overcome monitoring delays, predicted channel information is utilized. In this paper we are analyzing the outage probability of a max-min radio resource allocation (RRA) approach when Gaussian errors are present in the channel state information (CSI). As a communications channel, a Rayleigh fast fading channel is assumed. The outage probability is lower and upper bounded to get analytical performance approximations. Furthermore, the performance is also empirically evaluated by means of extensive Monte-Carlo simulations. The analysis shows, that the achievable outage probability mostly depends on the size of the resource pool and the quality of the CSI.