Theoretical models predict how paternal effort should vary depending on confidence of paternity and on the trade-offs between present and future reproduction. In this study we examine patterns of sperm precedence in Phyllomorpha laciniata and how confidence of paternity influences the willingness of males to carry eggs. Female golden egg bugs show a flexible pattern of oviposition behavior, which results in some eggs being carried by adults (mainly males) and some being laid on plants, where mortality rates are very high. Adults are more vulnerable to predators when carrying eggs; thus, it has been suggested that males should only accept eggs if there are chances that at least some of the eggs will be their true genetic offspring. We determined the confidence of paternity for naturally occurring individuals and its variation with the time. Paternity of eggs fertilized by the last males to mate with females previously mated in the field has been determined using amplified fragment length polymorphisms (AFLPs). The exclusion probability was 98%, showing that AFLP markers are suitable for paternity assignment. Sperm mixing seems the most likely mechanism of sperm competition, because the last male to copulate with field females sires an average of 43% of the eggs laid during the next five days. More importantly, the proportion of eggs sired does not change significantly during that period. We argue that intermediate levels of paternity can select for paternal care in this system because: (1) benefits of care in terms of offspring survival are very high; (2) males have nothing to gain from decreasing their parental effort in a given reproductive event because sperm mixing makes it difficult for males to reach high paternity levels and males are left with no cues to assess paternity; (3) males cannot chose to care for their offspring exclusively because they can neither discriminate their own eggs, nor can they predict when their own eggs will be produced; and (4) males suffer no loss of further matings with other females when they carry eggs. Thus, our findings do not support the traditional view that paternal investment is expected to arise only in species where confidence of paternity is high. The results suggest that females maximize the chances that several males will accept eggs at different times by promoting a mechanism of sperm mixing that ensures that all males that have copulated with a female have some chance of fathering offspring, that this probability remains constant with time, and that males have no cues as to when their own offspring will be produced.