||Climate change is a serious threat for agriculture, food security and the fight against poverty, especially in sub-Saharan Africa. With the IPCC climate outlook for the 21st century, the future of maize production in Kenya remains under threat of more intense and frequent droughts, fluctuations in temperature and more erratic rainfall patterns. Effective adaptation to these progressive changes in climatic condition is the key to securing food production and livelihoods for millions of poor people. This study was conducted to provide a comprehensive understanding of the adaptation behavior of maize-legume farm households in response to climate shocks, with emphasis on poverty, food security and gender perspectives. The research was conducted in rural Eastern and Western Kenya for the “Sustainable Intensification of Maize-Legume Farming Systems for Food Security in Eastern and Southern Africa (SIMLESA)” project. Data for this study are drawn from a SIMLESA baseline survey conducted in 2011. A total of 613 households were sampled by proportionate random sampling in two districts from the western region (Bungoma and Siaya) and three districts from the eastern region (Embu, Meru and Imenti South). Despite the area’s high potential for agricultural productivity, half of the surveyed households are living in poverty, with expenditure below 1 USD per capita per day. 70% of surveyed households in the western region compared to 30% in the eastern region identified themselves as food-insecure and almost 20% of households are female-headed in both regions. Between 1961 and 2012 the trend has been for annual rainfall to decrease in Siaya and Embu, to remain constant in Meru, and to increase in Bungoma. While Embu, Bungoma and Meru have experienced an increase of 1Cin the average daily temperature, in Siaya the temperature has decreased by 2°C. These observed changes in patterns of precipitation and temperature have put small-scale farm households in all districts under more pressure as regards production risks, as some crops, especially maize, are sensitive to certain temperature ranges. Between 2000 and 2010, almost all of the surveyed maize-legume farm households reported drought as the most frequent and severe climate shock on farm production, followed by crop pests and excessive rainfall. Drought, which on average occurred almost three times during the 10-year period, had the most severe effect on food-crop production and income when compared to crop pests and excessive rainfall. Based on farmers’ assessments, poor households with per capita expenditure in the lowest tertile were affected by drought, crop pests and excessive rainfall less frequently than households in the upper tertiles. With regard to food security, the effect of drought as measured by reduction in food production and income is more severe for food-secure households than for food insecure households, although the effect of crop pests on income is more severe for food-insecure households. Concerning gender, the adverse effect of crop pests on food production and income is found to be more severe for male-headed than for female-headed households. These results are contrary to what might be expected and should be treated with caution. One possible explanation may be related to the potential effect of having a large asset endowment and large-scale farm production, which makes the upper tertiles and food-secure households especially susceptible to climate shocks. A farm household’s adaptation to climate shocks follows a two-step process. As a response to the reduction in food production and income, a shock-affected household will first decide whether or not to undertake any action, before choosing a particular adaptation strategy from the available options. Almost 90% of shock-affected farmers applied at least one adaptation strategy, and most households reported that they had adapted to drought, with fewer reporting having adapted to crop pests and excessive rainfall. In the first step, results of probit regression show that the influence of previous frequent experience of drought significantly reduces the likelihood of adapting to drought and to excessive rainfall, while previous frequent experience of crop pests significantly increases the probability of adapting to all three types of shock. Although poverty and food insecurity are not found to have a significant influence on the adaptation decision for any shock, female-headed households are statistically 15% more inclined to adapt to excessive rainfall than male-headed households. The role of information through extension and membership of associations is found to support the decision to adapt to drought and excessive rainfall. The decision to adapt to drought and crop pests is found to increase with a proportion of steep farm land, but to decrease with farm size. Exposure to new technology is also found to influence the adaptation decision, as the adoption of improved maize varieties discourages crop pests, and maize-legume intercropping supports adaptation to excessive rainfall. However, education is shown to reduce the probability of adaptation to drought, while household size and credit access are found to reduce the probability of adaptation to crop pests. Distance to the main market and high rainfall variation show a significant positive effect on the probability of adapting to crop pests. High temperatures, on the other hand, tend to reduce the probability of adapting to drought and crop pests. Results from multivariate probit regression identify complementarity and substitutability between adaptation strategies. Among four common adaptation strategies for coping with drought, farm adjustment (e.g. use of improved seed varieties with early maturity and tolerance to stress, replanting, use of external inputs, conservation agricultural practices and crop diversification) is found to be a substitution strategy for selling assets and borrowing. Reducing consumption is found to be a substitution strategy for selling assets, but is complementary to borrowing for adapting to drought. To adapt to crop pests, results show farm adjustment to be a substitution strategy for selling assets and reducing consumption. However, selling assets is a complementary strategy for borrowing, but is a substitution strategy for seeking treatment for crop pests, which in turn, is a substitution strategy for borrowing. On the other hand, farm adjustment appears to be the single dominant adaptation strategy for coping with excessive rainfall. Regression results for all three types of shock further highlight the reinforcing influence between different types of shocks for households that have not only experienced one type of shock several times, but have also experienced multiple types of shock during the same period. The impact of a current shock may be aggravated by the impact of previous experience of shocks. As shown from adaptation to drought, for example, frequent experience of crop pests is found to support reducing consumption while discouraging farm adjustment and borrowing, while frequent experience of excessive rainfall is found to support farm adjustment but to discourage reducing consumption. Frequent experience of drought, on the other hand, encourages farm adjustment as an adaptation strategy for crop pests, while frequent experience of excessive rainfall reduces the likelihood of sales of assets and borrowing, but increases the probability of seeking treatment to cope with crop pests. Sampled households in the lowest tertile are less likely to reduce consumption and to borrow to adapt to drought, but they are more likely to adjust farm management to adapt to excessive rainfall. When faced with a reduction in food and income, food-insecure households are more inclined to eat less to adapt to drought, and to borrow to adapt to crop pests, but they are less likely to adjust farm management to adapt to excessive rainfall. Although there is no significant influence of gender on the choice of adaptation strategy for drought and excessive rainfall, female-headed households are found to be 15% more likely to sell assets than their male-headed counterparts to adapt to crop pests. Other socioeconomic variables, particularly household size, assets, off-farm income, membership of associations, contact with extension, and credit access, show diverse influence on choice of adaptation strategies. In terms of farm and farm-management variables, a small farm size encourages farm adjustment for adapting to drought, and seeking treatment for dealing with crop pests. A large area of steep slope on the farm is found to support adjustment in farm management for adapting to drought, but selling assets, reducing consumption and borrowing are more favored by households with a large area of flat land. Cultivation of improved maize varieties and a small holding of livestock further encourage farm adjustment for adapting to drought, while households with a large livestock holding are more likely to sell the animals. Households living in an area with high variations in rainfall and temperature are more likely to adjust their farm management as an adaptation strategy for drought. On the other hand, households living in low-temperature areas are more likely to sell assets to cope with drought and crop pests, whereas seeking treatment to cure crop pests is more likely for farm households in dry and cold areas. This study highlights the significant interdependency and reinforcing influence of frequency of other shock types on coping with and choosing an adaptation strategy for a particular shock. Therefore an effective policy scheme to support adaptation should not take any one shock in isolation, but should incorporate the context and composite implication of other shocks. Research into adapting to climate change and policy discussion should also recognize that a certain adaptation strategy can be applied for different types of shocks. For example, adjustment of farm technology and practices can be promoted to combat drought, crop pests and excessive rainfall. However, as farmers may choose to apply multiple adaptation strategies, it is important to take into account the complementarity and substitutability between different types of adaptation strategies, which is context-specific for each type of shock. Moreover, assistance should target food-insecure households belonging to the lower expenditure group and those led by female heads, as these are often disadvantaged in terms of asset ownership and access to the technology and information necessary for adaptation to shocks.