The main outstanding problem in our full understanding of the classical nova mechanism is the apparent discrepancy between mass transfer rates (10−9 to 10−8M⊙/yr) inferred from observations (Patterson, 1984, Ap.J.Suppl.231, 789), and those required by numerical models in order to reproduce nova characteristics (10−11 to 10−9M⊙/yr). The low accretion rates are needed in order to obtain powerful runaways and high values of Z in the ejecta by the diffusion-convection mechanism. The discrepancy seems to have sharpened by the realization that accretional heating (Shaviv and Starrfield, 1987, Ap.J.321, L51) and angular momentum transfer (Sparks and Kutter, 1987, Ap.J.321, 394 and Kutter and Sparks, 1987, Ap.J.321, 386) tend to lower the theoretical upper limit for Ṁ to about 10−10M⊙/yr. On the other hand, scenarios invoking variable mass accretion rates — hibernation (Shara et al, 1986, Ap.J.311, 163) or ‘mild’ hibernation (Livio, Shankar and Truran, 1988, Ap.J.330, 264) — have shown that the high observed rates immediately prior and following outbursts can be reconciled with lower average accretion rates over the period between outbursts.