Analytical data for nitrate and triazines from 566 samples collected over a 3-year period at Pleasant Grove Spring, Logan County, KY, were statistically analyzed to determine the minimum data set needed to calculate meaningful yearly averages for a conduit-flow karst spring. Results indicate that a biweekly sampling schedule augmented with bihourly samples from high-flow events will provide meaningful suspended-constituent and dissolved-constituent statistics. Unless collected over an extensive period of time, daily samples may not be representative and may also be autocorrelated. All high-flow events resulting in a significant deflection of a constituent from base-line concentrations should be sampled. Either the geometric mean or the flow-weighted average of the suspended constituents should be used. If automatic samplers are used, then they may be programmed to collect storm samples as frequently as every few minutes to provide details on the arrival time of constituents of interest. However, only samples collected bihourly should be used to calculate averages. By adopting a biweekly sampling schedule augmented with high-flow samples, the need to continuously monitor discharge, or to search for and analyze existing data to develop a statistically valid monitoring plan, is lessened.Analytical data for nitrate and triazines from 566 samples collected over a 3-year period at Pleasant Grove Spring, Logan County, KY, were statistically analyzed to determine the minimum data set needed to calculate meaningful yearly averages for a conduit-flow karst spring. Results indicate that a biweekly sampling schedule augmented with bihourly samples from high-flow events will provide meaningful suspended-constituent and dissolved-constituent statistics. Unless collected over an extensive period of time, daily samples may not be representative and may also be autocorrelated. All high-flow events resulting in a significant deflection of a constituent from base-line concentrations should be sampled. Either the geometric mean or the flow-weighted average of the suspended constituents should be used. If automatic samplers are used, then they may be programmed to collect storm samples as frequently as every few minutes to provide details on the arrival time of constituents of interest. However, only samples collected bihourly should be used to calculate averages. By adopting a biweekly sampling schedule augmented with high-flow samples, the need to continuously monitor discharge, or to search for and analyze existing data to develop a statistically valid monitoring plan, is lessened.