The net present value of costs and benefits from a pay-as-you-go social security system are negative for young people and positive for the elderly. If people all vote their financial selfinterest, there will be a pivotal age such that those who are younger favor smaller social security benefits and those who are older will favor larger benefits. For persons of each age and sex, we estimate the expected present value gained or lost from a small permanent increase in the amount of benefits, where the cost of these benefits is divided equally among the population of working age. Assuming that everyone votes his or her long run financial self-interest, and calculating the number of voters in the population of each age and sex, we can determine whether there is majority support for an increase or a decrease in social security benefits. We use statistics on the age distribution and mortality rates for the United States to explore the sensitivity of political support for social security to alternative assumptions about the discount rate, excess burden in taxation, voter participation rates, and birth, death, and migration rates. We find that a once-and-for-all decrease in benefits would be defeated by a majority of selfish voters under a wide range of parameters. We also study the predicted majority outcomes of votes on changing the retirement age.