This paper suggests a new and clear-cut way of deciding whether evolutionary effects are necessary to interpret the observed radio-source counts. To this end, a method is outlined for constructing the parent radio luminosity function from a complete sample of radio sources whose redshifts are all known. It is argued that if such a luminosity function can be constructed without evolu- tionary parameters, then evolutionary effects are not required. The method is illustrated by its applications to two well-known samples of radio sources. It is further shown that the non-evolving luminosity function satisfies the constraint of sky brightness and is consistent with the observed redshift-flux- density plot. Thus it appears from these investigations that evolution is not required either in luminosity or in number density.