One of the most important changes affecting the agricultural sector in the U.S. has been the increase in off-farm employment and multiple job-holding, especially among women on U.S. farms. This paper examines motivations for off-farm work among farm women in different farm production regions in the U.S. Further, the determinants of off-farm earnings of farm women (and their spouses) are analyzed as well as the receipt of employee benefits by either (or both) the farm woman and farm man. The paper goes beyond assessment of the important role of using off-farm work as a means of accessing health insurance and examines other types of benefits as well, including income for retirement. Background Based on a national survey of farm women conducted in 1980, Rosenfeld (1985) concluded that higher average education levels, advances in labor-saving technologies, and smaller family sizes contributed to 37 percent of U.S. farm women working in off-farm jobs at that time. Data from a recent national survey conducted by Pennsylvania State University showed that in the past two decades this percentage has increased to 52 percent of all farm women, and 62 percent of all working-age farm women. A recent study by Mishra, El-Osta, Morehart, Johnson and Hopkins (2002) concludes that around 71 percent of households in the United States have either the principal farm operator or spouse or both employed in off-farm jobs. Fuller and Madge (1976) observe that off-farm employment gives farm families a chance to interact with new people and to stabilize farm incomes. Mishra and Goodwin (1997) and Mishra (1996) found a positive correlation between off-farm employment and farm income variability, showing that off-farm employment helps many farm households to diversify their income risks. Further, one of the most important reasons for farm family members to work off the farm is to provide the family financial protection that is generally not economical for the farm business to purchase. These non-wage compensations include group health insurance, and group life insurance, as examples (Scholl, 1983; Jensen and Salant, 1985). Fringe benefits are important as they are not taxed as income and can be purchased by groups for lower per unit costs. Methodology Data. In 2001, a national survey of U.S. farm women was conducted by Pennsylvania State University in collaboration with researchers at the Economic Research Service (ERS, USDA) and in conjunction with the National Agricultural Statistics Service (NASS, USDA). The national survey was carried out by telephone. A total of 2,661 farm women responded to the survey that included questions on different motivations or reasons for off-farm work, off-farm wages and days worked, and receipt of employee benefits for both farm men and women. This information was collected in addition to data on characteristics of the farm household, the characteristics of the farm operation, and individual characteristics of both the farm woman and farm man, when present in the household. The data have been disaggregated by the nine production regions used by the USDA, and secondary data to reflect the effects of off-farm labor markets have been appended to the household dataset. Estimation Strategy. To assess those factors that affect the woman's reason for working off the farm, probit models are estimated. Responses range from not important, to somewhat important, to very important. Following this, models of the earnings of the farm man and farm woman in the household are estimated for the U.S. sample, using data on annual off-farm earnings. A Heckman-type approach is used for the earnings functions, since sample selectivity is likely to be a problem. In addition, the participation equations for men and women are jointly estimated, following Huffman (1991 ); Corsi and Findeis (2000 ); and Oluwole (2000). In the first stage, a bivariate probit model of participation in off-farm work is estimated and, if the work decisions between the farm man and women are shown to be jointly determined, the second stage estimation requires simultaneous estimation of the earnings functions of men and women. If sample selectivity is shown to be a problem, the inverse Mill's ratios are included in the second-stage model. Finally, data are available on whether the individual received employee benefits (overall and by type) from their off-farm employment. The following work choices are possible: no work, work in a part-time job with benefits, work in a part-time job without benefits, work in a full-time job with benefits, and work in a full-time job without benefits. A multinomial logit model is used to analyze the alternative work/benefit outcomes. The independent variables in the models include characteristics of the individual, the household, the farm and off-farm labor markets. The method of maximum likelihood is used for estimating the coefficients of the estimators. Preliminary Results Descriptive statistics show that working off the farm for benefits is an important or very important reason for the off-farm employment for many farm women in the United States. About one-third of women note that they work off the farm to help finance the farm operation. Further, preliminary models of earnings from off-farm work show that using a simultaneous equation model with corrections for sample selectivity is an appropriate approach. Earnings of U.S. farm woman are significantly affected by her level of education (as expected) and importantly by the characteristics of the off-farm labor market. The data also reveal that both farm women and men in the U.S. often receive employee benefits from their off-farm jobs, and that receipt of benefits tend to be in 'packages' - i.e., if the individual receives health insurance, they are also very likely to receive other benefits from their employer as well. The survey shows that among women with off-farm work, the following employee benefits from off-farm work were more common: health insurance (59%), life insurance (52%), a pension (54%), paid vacation leave (56%) and paid sick leave (58%). Among men with off-farm work, the following benefits were among the most commonly received: health insurance (67%), life insurance (58%), a retirement pension (59%), paid vacation leave (62%) and paid sick leave (53 %). Preliminary models again show that both education and labor market characteristics strongly affect benefit receipt and full-time job status, and that full-time work without benefits is more common than anticipated.