The aim of this study was to examine whether significant others can provide useful proxy information on the health-related quality of life (QL) of cancer patients. We examined the level and pattern of agreement between patient and proxy ratings of the EORTC QLQ-C30, the reliability and validity of both types of information, and the influence of several factors on the extent of agreement. QL ratings were obtained for 307 and 224 patient-proxy pairs (at baseline and follow-up, respectively). Agreement was moderate to good (ICC = 0.42 to 0.79). Multitrait-multimethod analysis showed good convergence and discrimination of specific QL domains. Comparison of mean scores revealed a small but systematic bias between patient and proxy ratings. The maximum level of disagreement was found at intermediate levels of QL, with smaller discrepancies noted for patients with either a relatively poor or good QL. Both patient and proxy QL ratings were reliable and responsive to changes over time. Several characteristics of the patients and their significant others were found to be associated with the level of agreement, but explained less than 15% of the variance in patient-proxy differences. In conclusion, the present findings lend support to the viability of employing significant others as proxy respondents of cancer patients' quality of life where this is necessary.