Respondents in developing countries frequently give a nonnumeric (Up-to-God) response to the desired family size question asked in fertility surveys. In literature nonnumeric respondents are viewed as fatalists or as those who consider family size to be beyond the realm of conscious choice and are usually excluded from analysis. This paper proposes and investigates an economic explanation for this response based on demand - supply model of fertility. Nonnumeric response is modeled as a discrete choice problem. The hypothesis is that respondents give nonnumeric response if their desired demand for children exceeds their potential supply of children. Data on husbands and wives from Bangladesh Demographic and Health Survey (1993-94) are used to assess the validity of the supply- constrained hypothesis. Reduced form probit model is set up to estimate the net effect of demand side and supply side variables on the likelihood of a spouse giving a nonnumeric response. Two purely supply related variables used in the analysis are marital duration and wife’s infecundity status. Joint husband-wife likelihood of giving nonnumeric response is estimated using bivariate probit which allows for within-couple correlation of unobserved variables. Results obtained from probit and bivariate probit models are consistent with expected effects. They show support for the supply-constrained view of nonnumeric response. Infecundity of wife significantly raises her own likelihood of giving nonnumeric response and marital duration significantly reduces husband’s likelihood of giving nonnumeric response. Husbands are more likely to give nonnumeric response than wives are. Also, some variables show systematic differences in the way they affect each spouse’s likelihood of giving nonnumeric response.