Time series that are observed neither regularly nor contemporaneously pose problems for most multivariate analyses. Common and intuitive solutions to these problems include interpolation and other types of imputation to a higher, regular frequency. However, interpolation is known to cause serious problems with the size and power of statistical tests. Due to the difficulty in dating paleoclimate data such as CO 2 concentrations and surface temperatures, time series of such measurements are observed neither regularly nor contemporaneously. This article presents large-and small-sample analyses of the size and power of cointegration tests of time series with these features and supports the robustness of cointegration of these two series found in the extant literature. Compared to linear or higher-order polynomial interpolation, step interpolation results in the least size distortion and is therefore recommended. IRREGULAR AND NON-CONTEMPORANEOUS SERIES 937 cliometrics, is provided by historical GDP data originally estimated by Angus Maddison, which extend back annually to 1950 for most countries, but irregularly as far back as 1 CE for a few countries.Paleoclimate data provide a fourth and striking example. Of particular importance are the series of carbon dioxide (CO 2 ) concentrations in parts per million by volume (ppm or ppmv) and temperatures in degrees Celsius ( • C). The era of human development and proliferation has occurred during the Holocene, which is a relatively small part (1-2%) of the paleoclimate record. Yet understanding the relationship between CO 2 and both local and global temperatures over the whole record can inform climate scientists, statisticians, and social scientists about the possible effects of future anthropogenically emitted CO 2 .These data are acquired by drilling ice cores, primarily in Antarctica or Greenland. Although the cores can be sampled at regular depths, dating the samples results in irregular spacing. When more than one ice core is sampled, the irregularity also means that observations of the relevant series are not contemporaneous. The process of collecting data from ice cores is discussed in more detail in Section 3.The principal statistical tool used to assess correlations of stochastically trending series over a long period -i.e., long-run comovement -is cointegration analysis. Cointegration analysis has been applied to paleoclimate data by Kaufmann and Juselius (2010a, 2010b, who find a cointegrating relationship between CO 2 concentrations and temperature. Katarina Juselius was one of the pioneers of likelihood-based analysis of cointegrated time series (Johansen and Juselius, 1990, 1992), while Robert Kaufmann was one of the first to apply cointegration analysis to recent climate series (Stern and Kaufmann, 2000;Kaufmann and Stern, 2002). To overcome irregularity and non-contemporaneity, these researchers linearly interpolated the paleoclimate series.The statistical literature on missing observations provides a wide range of alternatives to linear interpolation: both...