This paper describes a specific structured procedure for peer assisted learning in writing: Paired Writing (Topping, 1995a), and its effects on the quality of writing and attitudes toward writing in primary school children. Three action research projects are briefly reported. In the first, the method was adapted for use with five-year-old emergent writers, supported by eleven-year-old tutors. In the second, eight-year-old children worked either in cross-ability fixed-role pairs or in same-ability reciprocal-role pairs. Both groups gained, compared to control groups who wrote individually. However, positive effects for the more able helpers in cross-ability pairs could be delayed rather than immediate. In the third, Paired Writing operated in same-age cross-ability pairs, in a class of ten-year-olds with a high level of behaviour problems. The Paired Writing method proved practical and robust, adaptable to different classroom contexts, and beneficial in attitudes to writing and skill development. Its use within the literacy hour and beyond that across the curriculum is worth consideration by teachers seeking methods of proven effectiveness which can differentiate and enrich mechanistic prescriptions.