The salient features of a U-Pb concordia diagram (e.g., 206Pb/238U versus 207Pb/235U or 238U/206Pb versus 207Pb/206Pb) such as local modes and relative discordance are frequently obscured in diagrams based on large sets of data. Simplifying and plotting the data as univariate ages can aid interpretation, but the discordance filtering of such data is arbitrary. A new approach is proposed where the individual analyses are treated as bivariate normal distributions and their values are summed across an x-y grid to produce a three-dimensional probability distribution. This distribution can then be contoured and color mapped to illustrate features relevant to a concordia line. The mathematical basis for this approach is discussed along with the description of a package using the open-license R statistical package.