If you a have data only on the dependent variable, I suggest that you use ARIMA.
Even if you are building a traditional model with explanatory variables, which will be used for forecasting, I have found that it is a good idea to run the ARIMA procedure first. It gives you a pretty solid benchmark, to which you can compare the results of your traditional model to and can turn out to be a better forecasting tool than the model with a set of determining factors included.
If you would like some information on ARIMA, you can check the script, which I have co-authored on some econometric methods - copy is attached on RG.
I would presume that your data are seasonal and are stationary once the seasonal effects are removed. A SARMA or SARIMA might be appropriate or some form of regression with seasonal dummies.