The great majority of questionnaires measuring non-observable constructs such as attitude towards CALL are often developed from a specific point of view and are seldom followed by psychometrical validation. Psychometrical properties of the questionnaire, such as construct validity and reliability, then remain unanswered too often, laying a heavy burden upon the generalizability of the obtained results. In light of recent calls for more theory construction in CALL it would be very fruitful if researchers start to work with validated operationalizations of the variables they work with in developing CALL theories. The research presented in this paper aims at introducing a methodological approach to develop an empirically-based and psychometrically-sound instrument to measure the attitude towards computer-assisted learning (A-CAL), attitude towards foreign language learning (A-FLL) and, more specifically, attitude towards computer-assisted language learning (A-CALL). (Contains 9 tables and 3 figures.)