Without knowing what is in your model, the nature of your variables or how you constructed the model, it would be impossible to tell for certain. Is the coefficient for the interaction positive or negative? If it's positive then that merely means that one variable has a positive (or increasing) moderation or modification on the other variable in the interaction. A negative coefficient would be the reverse. Taken together, this is having some effect on your outcome of interest. It would be up to you to explain how and why one variable (and which one) is modifying the other. Literature may provide some guidance regarding the relationship between the variables in your interaction.

Let us illustrate this with a an example of everyday life. Imagine you run an anova, regression or whatever of happiness (0-100) and two factors, X1=having a left shoe (0=no,1=yes), X2=having a right shoe (0-1). My hint would be that happiness should not respond to changes in X1 that are unmatched by changes in X2. What is a single shoe good for? So in this case the meaning is obvious: you need both shoes to affect happiness.

In your case, whether the results make sense to you is ultimately a theoretical issue. You tested them independently and with the interaction, so you should have a theory for that. As @Joseph said, it is up to you to explain it, taking into account the lit.

Regards from Madrid

Plot you data (ideally in an interaction plot) and you will see what it means.

http://692r.blogspot.de/2013/02/interaction-plots.html

https://en.wikipedia.org/wiki/Interaction_(statistics)#Interaction_plots