The single most important environmental variable correlating with annual su
rvival of both juvenile and adult Great Tits Parus major is the beech crop
index (BCI). This index is a measure for the amount of seeds of beeches pre
sent in the winter, and correlates with crop size of several other tree spe
cies. Two, not mutually exclusive, hypotheses exist to explain the correlat
ion between BCI and annual survival. The first is that the amount of seeds
directly affects survival (winter-food limitation hypothesis). To test this
hypothesis we re-analysed a supplemental food experiment of Van Balen (198
0), extended with three more years of previously unpublished data. We found
that supplemental food increased survival of both juveniles and adults, ma
inly in low BCI years, confirming the winter-food limitation hypothesis. Th
e second hypothesis states that winters with a high BCI are preceded by spr
ings with high densities of caterpillars (the breeding-season food limitati
on hypothesis). Using data of three long term studies of Dutch Great Tit po
pulations we show that this hypothesis does not hold because (1) high BCI w
inters are not preceded by springs with high caterpillar densities, (2) ann
ual mean juvenile fledging mass increased with increasing caterpillar densi
ties, but no effect on parental mass was found, and (3) the annual survival
of juveniles was not affected by the annual mean fledging mass, nor was an
nual adult survival affected by their mean mass during chick rearing. Final
ly we show that, despite the fact that the level of BCI probably can be pre
dicted at the time of the breeding season and that both juvenile and adult
survival is affected by BCI, reproductive decisions (clutch size, laying da
te and percentage second broods) are not affected by the BCI levels in the
following winter.