Studies of the total spectra of planetary nebulae (PN) show that they usually emit significant fractions of their total energy at infrared wavelengths. Many PN have IRAS observations but we do not understand their broad-band colours. To model these observations requires a combination of a photoionization treatment of the ionized region and a radiative transfer treatment of any neutral dust shell. Including the effects of the central star evolution and an assumed galactic distribution allows many observables to be simulated. I find that the Schönberner 0.64 M⊙ evolutionary track is too slow in the transition to high temperature to match the IRAS colours of PN if a typical AGB wind expansion speed is 10 km s−1. Accelerating the evolution track by a factor of 2 produces reasonable agreement. An equal mixture of carbon-rich and oxygen-rich sources with mass loss rates of 2 × 10−5 M⊙ yr−1 seems to reasonably match the general properties of the PN data from IRAS although the details are still unclear. These models strongly indicate that the post-AGB luminosity decline deduced by Knapp (1986) from radio CO line observations is not correct.