OBJECTIVES: Prior to use in an international study involving children and adolescents suffering from schizophrenia or bipolar disorders, the original 21-item Caregiver Strain Questionnaire (CGSQ) underwent linguistic validation in 19 languages. The original scale, which measures the strain of caring for children with emotional and/or behavioural problems on their caregivers, was developed in US English. A rigorous methodology was required to ensure conceptual equivalence and cultural relevance across different languages. METHODS: The translation process was conducted by a specialist in each target country using the following standardized methodology: 1) two forward translations by professional translators who were native speakers of the target language and fluent in English; 2) comparison and reconciliation of the translations by the specialist in the target country and the translators; 3) backward translation by a native English speaker; 4) comparison of source and backward version; 5) review by a clinician; and 6) comprehension test on five parents of children suffering from schizophrenia or bipolar disorders. RESULTS: The translation process revealed two types of challenges: a) the translation of technical terms and b) the translation of idiomatic expressions. In the first case the expression "negative mental and physical health effects" required a simplified description in the translations. Similarly, the term "caregiver", having no literal equivalent, had to be replaced by the mention of the actual person providing care. As for translations of idiomatic expressions, paraphrases were often needed to convey the intended meaning. CONCLUSIONS: The 19 language versions of the CGSQ were established according to a rigorous translation methodology. The process aims to ensure conceptual equivalence across different language versions to facilitate international comparison and pooling of data. The linguistic validation process as a whole supports the advantage of integrating international feedback on concepts and wording before a questionnaire is finalized.