Mortality of patients (pts) affected with cancer can be partially unrelated to tumor, while accurate clinical investigation must be focused mainly on that tumor related. Specific survival (SS) is obtained by subtracting expected mortality (EM) – inferred from national or regional mortality tables – to the observed one. The expected survival (ES) is that enjoyed in the absence of tumor. ES corresponds to the time still at patients’ disposal in case of hypothetical full recovery and represents an absolute ceiling of maximal mean attainable SS for a given tumor in a given time. SS limits of variability clinically correspond to full recovery and death, and SS can be evaluated by the measure of how much, in a given time/site, a group of pts has approached the potentially attainable ES. On this full range of prognostic variability relies the comparability of SS, whatever time/site are involved. The number of years lost with respect to those expected represents a useful additional measure. ES and SS are subjected to the same conditioning factors, being contextualized into the evolving conditions of the health services, and depending on customs and behaviours. SS allowed comparison of the effectiveness of some chemotherapy regimens administered to pts with advanced-stage (IIB-IV) Hodgkin lymphoma (HL), whatever histologic subtype, age 16-65 years, of two randomized trials of the past decades. The IIL-HD9601 (1996-2000) enrolled 355 pts in three different arms: ABVD, Stanford-V and MOPPEBVCAD (MEC). The GISL-HD2000 (2000-2007) randomized 307 pts into three arms: ABVD (identical to IIL-HD9601), BEACOPP and COPPEBVCAD (CEC), a variant of MEC adopting cyclophosphamide instead of mechlorethamine at the dose considered with identical anti-lymphoma activity. CEC, ABVD and MEC showed the smallest ratio of expected/observed deaths. BEACOPP, ABVD and MEC allowed the relatively minimal loss of years of life per patient. The evaluation of the curves of ES confirmed the starting prognostic homogeneity of each arm of randomization (12-year ES from 0.965 [MEC] to 0.987 [BEACOPP]). The 12-year SS was 0.88, 0.81, 0.79, 0.79 for CEC, ABVD, BEACOPP, Stanford-V and MEC, respectively. The estimation of SS inferred from national or regional mortality tables allows a potentially unbiased comparison between treatments adopted in different time-spans, since it takes into account the social and health status along time. This method could be extended to other neoplasia.