I here propose a modification of the English butterfly census method (Pollard 1977) for studies of large areas with a relatively low density of butterflies. The transect line should not be more than 10 m from any point in open terrain and open forests, but only 5 m from sunny edges and patches with a high density of flowers. The transect line is adjusted between visits to most intensively cover sites with the highest density of flowers. However, between years the transect lines should be the same at a given locality and season to make statistical tests possible. Visits should only be made in suitable weather (English rules). A binocular with a short near limit facilitates species identification. The method was applied in 2001-2002 to a farm at Stenbrohult, southern Sweden. If visits are spaced about 14 days apart from the middle of May to the middle of August a high proportion of the resident species can be covered. Rare species may be missed even at weekly visits. However, from a statistically point of view it is better to cover two areas every 14th day than one area every 7th day. More spaced out visits minimize double counts of a given individual (statistical advantage) and also double the area that can be covered with a given effort.