This article outlines problems of implementation and clinical practice of randomised controlled trials in mental health services. Furthermore, it offers practical solutions taking into account the experiences with a randomisation process in a multisite EC-funded (EDEN-) study on the evaluation of acute treatment in psychiatric day hospitals. Identification of the problems follows the time-course of a research project: 1. Problems to be solved prior to the study's commencement: Definition of the eligibility criteria, information of clinically working colleagues. 2. Problems referring to the process of randomisation: Influence of clinical experience of the research fellows, precise time-point of implementing the randomisation into the process of admission, assessment of the patient's ability to give informed consent, patient's refusal of randomisation but agreement to study participation, availability of treatment places. 3. Problems which might occur after randomisation: Early break-off of treatment, transfer from one treatment setting to another. General conclusion: Detailed definitions of the randomisation procedure do not guarantee high performance quality and randomisation rates. Continuous precise assessment of the implementation into the clinical routines of every study centre, adaptation according to specific conditions and personal discussions with all participants are obligatory to establish and maintain a high quality of this important research procedure.