[en] The discrepancy between supernova explosion rates estimated from statistics of pulsars and of supernova remnants (SNRs) is well conciliated, if we consider that a considerable fraction of SNRs is missing. As the causes of missing SNRs, we have examined two cases, i.e., the supernova explosions have occurred in a hot, rarefied ambient matter or within stellar wind bubbles. For both cases, usual shell-forming radio SNRs are not expected at the stage, when their radii are smaller than --30 pc. From these, we predict the supernova rate would be (2 -- 5) x r sub(SN)sup(ob), with r sub(SN)sup(ob) being the rate determined from the counts of radio SNRs. This rate is not inconsistent with the one determined from pulsar statistics. Some other problems on missing SNRs will also be discussed. (author)