Water　distribution　network　optimisation　is　highly　complex　and　possesses　non-linear　relationship　of　the　problem　variables,　cascading　nature　of　hydraulic　network　that　make　the　problem　of　finding　global　optimum　difficulty　using　standard　optimisation　methods.　An　improved　particle　swarm　optimisation　(IPSO)　algorithm　is　presented　to　overcome　the　drawback　that　the　basic　particle　swarm　optimisation　(PSO)　algorithm　is　easy　to　fall　into　local　optimal　solutions　in　the　optimal　design　of　water　distribution　network.　The　procedure　of　optimisation　is　divided　into　two　phases　and　searched　by　PSO　with　extremum　disturbed　arithmetic　operator　and　differential　evolution　(DE)　respectively.　Extremum　disturbed　PSO　accelerates　the　particles　to　overstep　the　local　extremum　and　DE　keeps　the　diversity　of　population,　with　two　swarms　exchanging　information　in　each　iteration　to　avoid　local　optimum　and　reduce　iterations.　Water　distribution　networks　are　designed　by　improved　particle　swarm　algorithm　to　achieve　the　goal　of　lowest　cost　of　network.　Application　results　show　that　IPSO　reduces　the　cost　of　network　with　rapid　convergence　rate　and　strong　global　searching　ability.　Copyright　©　2011　Inderscience　Enterprises　Ltd.