Macroeconomic variables exhibit fluctuations around their long-term trends. Contemporary analysis generally separates growth trends and business cycles. Real business cycle, new Keynesian, and even Kaleckian theories which assume stable steady growth paths explain cycles through stochastic shocks whose effects on output and employment are then mediated by some propagation mechanism. In contrast, this paper provides a unified Keynesian framework of cycle and trend: we focus on Kaldorian and Robinsonian models in which the average utilization rate tends to fluctuate around a desired level that is largely independent of aggregate demand, and in which the fluctuations arise endogenously from the existence of feedback effects that make the balanced growth path locally unstable. Kaldorian and Robinsonian models have similar steady growth characteristics but differ with respect to underlying short-run dynamics. When firms find themselves with more excess capacity than they think optimal, they will either reduce the scale of their investment plans-the approach stressed by Harrod and incorporated into our Kaldorian model-or they will reduce their markup in order to increase output-the mechanism emphasized by Robinson. Another difference concerns relative adjustment speeds: Kaldorian models view price adjustment as fast (er than output adjustment); Robinsonian models reverse this assumption and treat output as the fast variable. We(i) analyze the implications of these differences for the cyclical properties of the economy and (ii) evaluate the consistency of the theoretical predictions with empirical evidence for the US. Using simple prototypes of the Kaldorian and Robinsonian position, our simulations of both models successfully reproduce some of the key patterns in the US data. This is a relatively weak test, however, and regressions fail to support Robinsonian specifications of changes in the profit share. The estimation of a Kaldorian output adjustment mechanism, by contrast, produces plausible coefficients. The estimated investment function is consistent with both the Kaldorian and Robinsonian models but violates the assumptions underlying the Kaleckian growth model: the long-run sensitivity of accumulation to changes in the utilization rate far exceeds the limits imposed by a Kaleckian stability condition. Overall, the empirical evidence gives tentative support to the Kaldorian approach.