We consider the problem of distributed task allocation across swarm units comprising a swarm-based system. We model task allocation as a dynamic pricing problem within a market setup. Experimental results within a simulated environment for a distributed automatic target recognition application show that the market-based task selection mechanism compares favorably with other heuristic-based task selection strategies.