The study demonstrates how unobserved component modeling, also known as structural time series modeling, can be usefully applied to forecast non-farm employment for the Nash-ville MSA. Short-term out-of-sample forecasts are provided for total employment and its three components: services, construction, and manufacturing. The forecasts are compared to those of a simple vector autoregression. It is shown that the suggested methodology provides very ac-curate short-term forecasts even in the absence of a full set of independent regressors. In addition, it makes it possible to back out long-term trends, which aid the forecaster in making long-term projections of sectoral employment.