The general problems of conducting research among members of leftist parties are described, with particular reference to the parties on the Brazilian Left. The comparative advantages of Q methodology are outlined in terms of the selection of statements and respondents, and the modifications which have to be effected under shifting circumstances.