Level crossing rate (LCR) is a well-known statistical tool that is related to the duration of a random stationary fading process \emph{on average}. In doing so, LCR cannot capture the behavior of \emph{extremely rare} random events. Nonetheless, the latter events play a key role in the performance of ultra-reliable and low-latency communication systems rather than their average (expectation) counterparts. In this paper, for the first time, we extend the notion of LCR to address this issue and sufficiently characterize the statistical behavior of extreme maxima or minima. This new indicator, entitled as extreme LCR (ELCR), is analytically introduced and evaluated by resorting to the extreme value theory and risk assessment. Capitalizing on ELCR, some key performance metrics emerge, i.e., the maximum outage duration, minimum effective duration, maximum packet error rate, and maximum transmission delay. They are all derived in simple closed-form expressions. The theoretical results are cross-compared and verified via extensive simulations whereas some useful engineering insights are manifested.