Following a previous systematic study involving calculations of evolut
ionary sequences of nova outbursts through several cycles, for combina
tions of parameters-the accreting white dwarf (WD's) mass, its core te
mperature, and the mass transfer rate-spanning the entire parameter sp
ace, assuming CO WDs (with C and O in equal mass fractions), we now co
nsider the detailed composition of the ejecta for the subset of models
which simulated classical nova outburst. We also investigate the effe
ct of the additional input parameter-the WD composition-on nova charac
teristics by calculating evolutionary sequences with pure-carbon and p
ure-oxygen WD progenitors. The stellar evolution code used includes an
extended nuclear reactions network, OPAL opacities, and diffusion of
all elements. Our main conclusions are that CO progenitors reproduce m
ost of the observed abundances and abundance ratios, and that correlat
ions between them, if any, are in very good agreement with observation
s. The WD composition is generally not reflected in the abundances of
the ejecta: whereas a large fraction of the carbon is always turned in
to nitrogen, oxygen is in some cases unaffected and in others almost c
ompletely destroyed. Hence ejecta abundances cannot be used to deduce
the WD composition. Ejected masses of pure O WD progenitors exceed tho
se of CO progenitors, sometimes by a factor of 4. The isotopes C-13 an
d O-17 are in all cases significantly overabundant, compared to the so
lar composition: C-12/C-13 varies between 0.97 and 3.8 (by number) and
O-16/O-17 varies between 1.8 and 55; N-15, however, is sometimes grea
tly enhanced and sometimes underabundant, N-14/N-15 varying over a ver
y wide range, from 2.4 to 33,000.