Information retrieval systems are complex in nature due to the interactions of document, query, and matching subsystems involved in the process of retrieval. Researchers have applied probabilistic, knowledge-based, and, more recently, artificial intelligence based techniques like neural networks and symbolic learning to this problem. Very few researchers have tried to use evolutionary algorithms like genetic algorithms (GA's). Previous attempts at using GA's have concentrated on modifying the document representations or modifying the query representations. In this research, we explore the possibility of applying GA's to adapt the matching functions used in retrieval. We have described a method where an overall matching function is achieved by combining the results of the individual matching functions. The weights associated with individual matching functions have been adapted using GA's. We tested the method on two document collections. Experiments on these collections suggest that a GA based matching function adaptation significantly improves retrieval performance compared to the performance obtained by the best individual matching function. We believe the promising outcomes of the GA based matching function adaptation merits continuing research. We briefly present possible areas of future research such as simultaneous adaptations of the three subsystems involved in retrieval, user profiling using this approach, and evolving new matching functions.