Recent moves in the European Union have beenmade towards a toughening of legislation onbathing water quality. This has focussedpolicy-makers thoughts on the welfare benefitsresulting from such improvements, especiallygiven their cost. Our paper uses a combinedstated and revealed preference approach tovalue coastal water quality improvements,focussing on an area of Scotland which hasconsistently failed to meet standards under theBathing Waters Directive. We combine data onreal behaviour with data on contingentbehaviour using a random effects negativebinomial panel model. This allows us to predictboth the change in participation (trips) shouldwater quality be improved, and the welfareincrease per trip. Our model includes allowancefor the existence of substitute sites, and forchanges in recreational behaviour during abeach visit.