The risk of giving birth to a low birth weight baby is known to be associated with poor material circumstances during the mother's own childhood. In addition to this long-term effect, an association is also apparent of low birth weight with the mother's current social class, measured in terms of her husband's occupation. At least two interpretations of this are possible: a true short-term effect, and/or selective social mobility (upward or downward). According to the latter hypothesis, social class at marriage reflects the operation of selective social processes such that taller and better educated women tend to marry men of higher social class: because both attributes are negatively associated with the risk of low birth weight, a short-term effect is mimicked. This paper investigates the strength of this effect, using data from a longitudinal study, the National Child Development Study. The possibility is also explored that the social class gradients at different ages are not independent: for example that a beneficial socio-economic environment in early life can compensate for hardship later on, in terms of the risk of low birth weight, and vice versa. A social class gradient was observed in the proportion of low birth weight deliveries, both at the time of the respondent's own birth and at the time of her marriage. Depending on the assumption made concerning the relationship of height with low birth weight, selective mobility for height explained 10.7% or 16.3% of the apparent short-term gradient in low birth weight. Selective mobility for educational level did not have any effect. Further analyses suggested that having belonged to a higher social class either in early childhood or at marriage had a beneficial effect, notwithstanding the direction of any mobility experienced. The social class of the woman's father when she was 16 was not associated with the low birth weight rate, and upward mobility during the respondent's childhood appeared to carry an increased risk.