Application of time-lapse seismic data to reservoir monitoring and/or carbon dioxide sequestration programs still faces challenges arising from data repeatability and routine processing practices. Emphasis has therefore been placed on structurally imaging the subsurface rather than preserving seismic wavelet characteristics. Our need for high-resolution seismic data and wavelet-oriented processing required attention to detail coupled with a non-differencing approach to interpretation to successfully monitor a weak time-lapse anomaly in a carbonate reservoir. Near-simultaneous processing flow for baseline and monitor data sets with special attention paid to preserving and enhancing monitor seismic wavelet characteristics and adopting an interpretation approach that would mitigate effects of less-than-perfect cross-equalization enabled us to detect and monitor a very weak (~10%) CO2 flood anomaly. Conservative cross-equalization, though desired for preserving weak TL anomalies, renders data differencing unsuitable for most subtle time-lapse anomalies. Higher-resolution data improves the signal's sensitivity to fluid effects on time-lapse seismic and lessens the risk associated with monitoring thin and inhomogeneous carbonate reservoirs. However, high-resolution data is more susceptible to repeatability errors during differencing. Improvement in detectability of change using our non-differencing approach stems from enhanced emphasis on data quality and repeatability, use of areal textures on seismic horizon attribute slices, avoiding the noise amplification from differencing, and textural changes of time-lapse signatures from weak fluid effects.