Cattle exhibit a circadian rhythm of grazing and lying behaviour. Understanding how available forage (FA) interrupts this rhythm and at what time or spans of time throughout the day could be beneficial in understanding grazing dynamics. Predictable changes in activity using FA would also permit changes in activity to indicate when forage supply is approaching a threshold for optimal herd productivity. The objective of the present study was to utilise wavelet techniques to evaluate the changes in herd-level lying, grazing and standing or walking behaviour in response to varying levels of FA. This objective was accomplished by (1) applying the discrete Haar wavelet transformation to lying and grazing activity, (2) examining the relationship of the transformation coefficients with FA, and (3) developing coefficient-prediction equations to model the relationship between the coefficient and FA. Data for the study were established by monitoring the grazing behaviour of up to three cattle herds, during June, July and August of Years 2010, 2011 and 2012. The minimal FA studied was 12.8 kg DM/100 kg bodyweight and maximum FA was 56.5 kg DM/100 kg bodyweight. Lying and grazing activity were significantly (P < 0.05) correlated with a single-wavelet coefficient and these occurred at different scales (timespans). Linear models to predict coefficients were constructed for correlations of P < 0.05. The coefficient for lying was predicted as 0.011944 × FA – 0.641426 (r2 = 0.31), and the coefficient for grazing was predicted as 0.009547 × FA – 0.847766 (r2 = 0.21). On the basis of location of the predicted coefficients within the wavelet pyramid, from 0600 hours to 0900 hours, lying activity increased as FA increased; however, from 1000 hours to 1300 hours, lying activity decreased as FA increased. Lying activity converged during the timespan from 1400 hours to 2100 hours. Prior to 1300 hours, fewer cattle were predicted grazing as FA decreased. Following 1300 hours, the proportion of cattle grazing increased as FA decreased. In conclusion, the use of wavelet techniques in combination with linear regression provided a mechanism to study the magnitude of changes in herd grazing and lying activities at different scales of time throughout the day due to varying quantities of FA. However, the slope coefficient of the linear models to estimate the wavelet coefficients that were significantly correlated with FA resulted in modest changes in estimated activity. Therefore, utilising time-of-day-related changes in activity may not be a practical mechanism to delineate modest forage shortages.