The system BSD 2000 has been in clinical use for regional hyperthermia for more than 10 years. Several technical details of this hyperthermia system, as well as the results of clinical studies employing this system have been investigated. The intention of this paper is to investigate the correlation between technical efficiency or feasibility of hyperthermia with the BSD 2000, in terms of power densities and temperatures depending upon parameters such as tumour histology, tumour location, patient age, patient sex, and patient cross section. The possible conclusions of predictive factors derived from the above correlations were closely scrutinized. Data acquired from 772 treatment sessions of 190 patients with pelvic tumours, mainly sarcomas and carcinomas of the rectum, cervix, prostate and anus, have been evaluated. For every session, index temperatures T90 (temperature attained at 90% of tumour related measurement points), cumulative minutes for T90 > Tref, tumour related power density (SAR: specific absorption rate, in W/kg) and the effective perfusion Weff (in ml/100 g min) were calculated. Temperatures were measured either invasively or endoluminally. The statistics software SPSS was employed subsequently for univariate, as well as multivariate analyses. The results exhibit that index temperatures mainly depend on the power density SAR and the hyperthermia induced effective perfusion. The total power P (in 100 W) and, complementarily, the relative power density absolute value(SAR) (= SAR/P) seem to have lesser influence. Clear differences between the tumour entities were established regarding their index temperatures and temperature distributions. SAR, Weff and P were correlated with several anatomical, biological and clinical factors. Sessions rendering low index temperatures and SAR values also revealed decreased individual tolerance to the treatment. This clearly displays that power-induced side effects define the limits of the efficiency of regional hyperthermia. Equivalent relationships and correlations are derived from intratumoural and endoluminal thermometry. Individual limitations of regional hyperthermia caused by anatomical, biological and clinical factors are liable to be difficult to overcome with the rather restricted potentials of the BSD 2000 system to control the SAR distribution.