Purpose – The purpose of this paper is to study empirically the dynamics of Turkish bilateral trade between Turkey and her nine trading partners, in addition to aggregate trade balance data. Design/methodology/approach – The paper employs cointegration, generalized impulse response analysis, and stability tests. Findings – The empirical results suggest non-existence of the J-curve effect at disaggregate and aggregate levels. However, Marshall-Lerner condition holds for the aggregate data along with some of the trading partners. With regard to the stability of trade balance equations, the findings are mixed. Practical implications – Conclusions drawn from this study could be useful for the policy makers of governments and practitioners in international trade organizations. Originality/value – This paper extends the existing literature by providing initial evidence at disaggregate data in the case of Turkey. Moreover, for the first time disaggregate and aggregate data are utilized in the same analysis.