A number of different approaches to likelihood inference for models specified by families of unnormalized densities. The main approach used here will be the Monte Carlo likelihood (MCL) of Geyer and Thompson (1992) and Geyer (1994). This will be compared with the Monte Carlo Newton-Raphson (MCNR) approach of Penttinen (1984) and the stochastic approximation approach of Younes (1988) and Moyeed and Baddeley (1991). For other approaches, see the introduction and references of Geyer and Thompson (1992).