This paper addresses the issue of how regions grow and decline, focusing on actual productive capacity within a region (rather than population, income, or migration). Traditional theory has emphasized the role of industrial inertia and cumulative causation in imposing stability on the spatial distribution of capital stock over time, while recent theories emanating from the catastrophe paradigm attempt to demonstrate the capacity for rapid and dramatic reversal of relative growth trends within regional systems. This paper offers an empirical analysis of capital accumulation in U.S. regions over the period 1954-1976 for total manufacturing and four 2-digit industries. Its aim is to assess the degree of stability or volatility which characterized the actual regional patterns of capital over this period. To achieve this goal, the analysis first assesses capital growth rates in different regions over time as well as the changing distribution of capital stock across regions. Second, the issue of volatility is directly addressed through the time-series modelling of regional capital formation in individual regions. It is concluded that, while considerable volatility is apparent in individual regions and industries, these changes are far from catastrophic in character. An alternative conceptualization of the regional investment process is offered which explains long-run change as the accumulation of short-run adjustments to capital.