The existing macroeconometric evidence lends support to the wage Phillips curve by showing a negative relation between the rate of change in wages and the unemployment rate, conditional on lagged price inflation. Most theoretical models of wage setting, however, generate a "wage curve," described by a negative relation between the level of the real wage and unemployment. Real wage dynamics have important implications for how shocks affect aggregate consumer price inflation, and for the determination of the natural rate of unemployment. This paper examines the dynamics of the aggregate wages in Canada, and tests whether real wages and unemployment have a long-term level relationship. The results indicate that a simple aggregate wage Phillips curve continues to describe the behaviour of aggregate wages in Canada quite well. The micro evidence, however, does not unequivocally support one specification against the other; rather, what seems to emerge is more complex wage dynamics better described in an error-correction specification. Wage changes reflect the short-run movement in the unemployment rate, while they adjust towards a long-run equilibrium level, as could be described in a wage curve model.