Efficient channel allocation to mobile hosts is of utmost importance in a cellular network. A genetic algorithm (GA), which is a useful tool in solving optimization problems, is explored to design a fault-tolerant cellular channel allocation model that allows a cell to continue communicating with its mobile hosts, even if there are insufficient channels available in the cell. Sometimes, the load over a cell may increase to the extent that it needs more channels than it actually has in order to handle the traffic. On the other hand, it is quite possible that the load in some other cell is less than its channel capacity, resulting in underutilization of the channels. This problem is solved by temporarily taking unutilized channels from cells that have lesser load and allocating them to the cells that are overloaded. We propose a model that reuses available channels more efficiently. The model also considers the handoff by using the reserved channel technique. A reserved pool of channels makes the model fault tolerant. Thus, the proposed work uses GA for fault-tolerant dynamic channel allocation to minimize the average number of blocked hosts and handoff failures in the mobile computing network. Simulation experiments evaluate the performance of the proposed model. Comparison of the results with the two recent earlier models reveals that the proposed model works better in serving mobile hosts.