Distribution based artificial fish swarm is a new heuristic for continuous global optimization. Based on the artificial fish swarm paradigm, the new algorithm generates trial points from the Gaussian distribution, where the mean is the target point and the standard deviation is the difference between the current and the target point. A local search procedure is incorporated into the algorithm aiming to improve the quality of the solutions. The adopted approach for handling the constraints of problem relies on a simple heuristic that uses feasibility and dominance rules. A comparison with a previous version, where the mean of the Gaussian distribution is the midpoint between the current and the target point, is investigated using a set of engineering design problems.