Markov models of loss incidents happening during Packet Voice Communications , a.k.a. Markov loss models, are needful for many engineering tasks, namely network dimensioning and automatic quality assessment. Two very simple ones are Bernoulli and 2-state Markov models, but they carry limited information about incurred loss incidents. On the other hand, a general Markov loss model including 2 k states, where k is the window length used for observing the voice packet arrival process is characterized by an intractable modeling complexity and an excessive lookahead delay. Moreover, legacy Markov loss models concentrate mostly on capturing some physical characteristics of loss incidents, rather than their perceived effects. This paper proposes a comprehensive and rather detailed Markov loss model considering the distinguished perceived effects caused by different loss incidents. Specifically, it explicitly differentiates between (1) isolated 20 msec loss incidents which are inaudible by the human ears, (2) highly and lowly frequent short loss incidents (20-80 msec) that are perceived by humans as bubbles and (3) long loss incidents (≥ 80 msec) inducing in-* terruptions that dramatically decrease speech intelligibility. Our numerical analysis show that our Markov loss model captures subtle characteristics of loss incidents observed in empirical traces sampled over representative network paths.