Roads and bridges are designed with defined useful lives but the extent to which they last, as Scott notes, depends hugely on individual circumstances. Some roads don't last as long as their design lives while some outlast them several times over.
I don't have specific information on US practice, but in Britain design codes currently in use call for a design life of 40 years for the road and 120 years for the bridge. There is no clear economic argument in favor of using one-size-fits-all design lives for highway infrastructure, but I suspect that in most cases, in most US states, the nominal design life for road bridges is significantly longer than that used for the roadbed. This is because when a road fails, the main consequences are primarily economic, while if a bridge fails, people can lose their lives.
A key consideration, when choosing a design life for a piece of highway infrastructure in a corridor which is expected to remain in use well beyond any reasonable choice of design life for the actual physical fabric, is rollover costs. What will need to be done when the highway comes to the end of its natural life and how much will that cost? In the case of rural Interstates which are considered unlikely to need widening ever, for example, it is often economic to design for 40 years or even longer, while for a highway which is likely to need widening within a few decades, a shorter design life of 20 to 25 years may be more appropriate. This is where sensitivity testing has to be done as part of the cost-benefit analysis--the owning agency has to know how the results change for various possible choices of design life. Frequently agencies also compute costs and benefits for diverging scenarios of traffic growth. In Spain, for example, proyectos de construcción report design-year traffic estimates in terms of low growth, medium growth, and high growth, and report design-year LOS for each scenario.