Metamerism is an effect where two different pigments look the same under a particular light. This occurs because the light that reaches our eye is effectively the product of the spectrum of the light multiplied by the reflectivity of the pigment at each wavelength. To get to our perception of color, we also multiply by the sensitivity of each cone (red, blue, green) in our eye and integrate over the whole spectrum. The resulting triple (one integration for each type of protein in the cones of our eye) defines the color we see. If two pigments share the same triple, they look the same. However, if viewed in a different light, with a different spectra, this integration may not come out the same. This causes the pigments to appear as different colors under this light.
CRI is one model which attempts to capture this effect. While the "color" of light boils the light's spectrum into a single number, the CRI measures the color of several swatches under the light and reports on the differences between the colors of those swatches and the color they would be under a perfect black-body light of the same "color. Its similar to the concepts of average and standard deviation in statistics. The color is similar to the idea of an average, trying to capture the overall color of the light, while the CRI is trying to capture something similar to the standard deviation, reporting how much variance there is in color with respect to this ideal average light.
A CRI of 100 is identical to a blackbody light (i.e. daylight or an incandescent light). The further you fall off from 100, the more color variations may appear. A CRI of 90 may have lots of colors that appear slightly off, or it may perfectly reproduce most colors and have one color that is very far off.
For two lights to have the same color and different CRI, they must have different spectra that, when you apply the equations to determine color temperature, yield the same temperature.