Using the augmented version of the Blanchard - Perotti’s SVAR technique, this paper seeks to empirically estimate the size of fiscal multipliers in Turkey over the period 2002:q3-2016:q2. In contrast to many previous papers that concentrate on fiscal policy instruments -taxes and government spending- at the aggregate level, in the paper we consider these instruments at the sub-component level. We examine output responses to discretionary changes in five fiscal variables (value-added tax, special consumption tax, personal income tax, real government spending, and transfer payments), and then we estimate the size of fiscal multipliers for taxes and government spending. Overall, our empirical findings indicate that the size of multipliers for taxes is different from that of government spending. Depending on the sub-components, the size of the multiplier ranges from -0.83 to -0.27 for taxes, and from 0.02 to 0.98 for government spending respectively. Overall, these findings corroborate the idea that a shock to government spending creates a (weak) Keynesian effect on GDP in the short run, while a shock to taxes brings about a non-Keynesian effect.