Noise emissions from construction sites are inherently unsteady. Noise emissions vary due to many causes, including the noise sources frequently changing in location, orientation, the types of activities they perform, and the acoustic shielding due to structures and/or terrain. The noise that arrives at receivers from construction site equipment can fluctuate over all time scales, from seconds to hours, days, months or years. Prediction of noise levels typically assumes either a worst-case approach in which all noise sources are assumed to be operating simultaneously, or by predicting an energy-average (Leq) level over a long time period. In the latter case, an energy-average (Leq) noise level is predicted at receivers, based on the anticipated percentage utilisation of the various noise sources on the construction site that is, the fraction of time that each item of equipment is operating or not during the averaging time period. This paper presents a method that may be used to estimate the variability of noise emissions from the site and the corresponding noise immissions at receivers using the Monte-Carlo simulation method. Using this method, the expected minimum, maximum, percentiles and energy-average (Leq) noise immission levels at receivers can be predicted.