From the definition of fuzzy derivative and integral via Zadeh's extension of the derivative and integral for classical functions we obtain a fundamental theorem of calculus and develop a new theory for fuzzy differential equations (FDEs). Different from the previous concepts of fuzzy derivatives (Hukuhara and generalized derivatives) and integrals, defined for fuzzy-set-valued functions, the approach we propose deals with fuzzy bunches of functions (fuzzy subsets of spaces of functions). Under reasonable conditions, the new operations are equivalent to differentiating (or integrating) the classical functions of the levels.We present the most known previous approaches of FDEs. Comparisons with the new theory we propose are carried out calculating fuzzy attainable sets of the solutions. Under certain conditions, the solutions via strongly generalized derivative coincide with solutions using our approach. The same happens with solutions to fuzzy differential inclusions and Zadeh's extension of the crisp solution. Although these two methods do not treat FDEs, they are widespread for making use of classical functions (similarly to what is proposed in this thesis) and for preserving properties of classical dynamical systems. These are advantageous features since it shows that the new theory presents desirable properties of the other two mentioned theories (allowing for instance periodicity and stability of solutions), besides treating FDEs.The theory is illustrated by applying it on biological models and commenting the results.