Rural households in sub-Saharan Africa face many risks and are therefore often vulnerable to poverty. Income diversification has been proposed as a potentially effective strategy to mitigate their risks. This paper explores how portfolio decisions of fishery-dependent households in Cameroon affect income and risk in different production systems. Data on income variation across time were elicited using the visual impact method. Portfolio theory and stochastic dominance rules were used to derive and compare income distributions. The results show a considerable variation in incomes despite diversification into crops or fishing. We conclude that any combination of cropping (sorghum, millet, rice) and fishing is not effective in reducing risk as long as the dependency on climate remains strong (resulting in high covariance of production output). We show that development intervention strategies that aim specifically at changing the co-variation structure of income flows are the most successful in reducing risk and potentially increasing income.