The provincial government of Prince Edward Island, Canada, required timely, objective, and accurate annual crop area statistics and mapping for 2006 to 2008. Consequently, Statistics Canada conducted a survey incorporating medium- resolution satellite imagery (10 to 30 m) and statistical survey methods. The objective was to produce crop area estimates with a coefficient of variation (CV) as a measure of accuracy, and to produce maps showing the distribution and location of different crops and land cover types. Conducting a multi-year study gave Statistics Canada an opportunity to refine the methodology in order to reduce data collection and processing while maintaining high quality estimates. The optimal parameters to design the area survey and the choice of the satellite imagery depend on a number of parameters: average size of field, desired accuracy, distribution and abundance of crop types, as well as availability of historical data. Data collection, using a sample frame of the entire province, was conducted from the roadside or air, so did not require any input from farmers. Results were produced within one month after data collection. When comparing 2006 with 2008, accuracy was maintained for potato (CV of 1.6% in 2006, 1.9% in 2008) and total agriculture area (CV of 2.7% in 2006, 2.3% in 2008), while the amount of sampled land surveyed dropped from 16% to 7% of the total area (5,700 km2), substantially reducing the amount of work for preparation, collection and processing. The results of this study will help users to select optimal parameters for an inventory of their area of interest. The benefits of using this type of survey compared to the traditional crop survey methods are: relatively low cost, no response burden on farmers, timeliness, objectivity and accuracy of the estimates, and no sophisticated survey infrastructure requirement.