Patient-centredness has been shown to be associated with improved patient outcomes in the West. The objectives of this study were: (i) to further test a specific method for measuring patient-centredness that had previously demonstrated validity and reliability, and (ii) to test the effectiveness of a patient-centred approach amongst poor, non-Western people in South Africa. Patient-centredness was measured in terms of the practitioner's facilitation of the patient's reasons for coming, including symptoms, thoughts, feelings, and expectations. The study involved nurse practitioners and medical doctors in three primary care settings with patients from eight language groups. The method for measuring patient-centredness was found to be valid, reliable (inter-rater correlations, rs = 0.95, 0.88, and 0.87), sensitive, and practical, being inexpensive, time efficient, suitable for consultations involving interpreters, and not requiring transcripts. The score for the first 2 minutes of the consultation correlated highly with the score for the entire consultation (rs = 0.92), which could make the method useful on a large scale. Patient-centredness itself, was also time effective, applicable to cross-cultural consultations involving interpreters, and was associated with patients feeling understood (P = 0.03), patient-practitioner agreement (P = 0.049), symptom resolution (P = 0.01), and concern resolution (P = 0.006). This study supports the effectiveness of patient-centred interviewing in a non-Western setting as well as this method of assessing it.