Depends on the strain range. For ABS I think it's better to use nonlinear analysis since its elastic region is very narrow. In ANSYS you can try Multilinear Kinematic Hardening model for the mechanical property of the material.

ABS and absolute majority of plastics are non-linear (in deformational sense) materials. 

As collegue Kan Wang said, linear region of stress-strain curve is small. There not essential difficulties to describe this nonlinearity by step-wise functions available in FEA software.    

I'd start with linear. Fast results, lets you decide on the next steps.

Depending on what is non-linear in your system (Boundary conditions, material, time dependent properties etc.), you might want to check out the literature and decide how much of an error you get from linear analysis. If it is too great, go non-linear.

I agree with what was previously said about plastics being non linear but for small deformations you should be ok. 

If you are trying to observe failure in the part, small deformation assumption is null and void.

Simple google search will yield some nice results. I found, for example, a very nice document by Dr. Vladimír Ivančo (Technical University of Kosice, Slovakia).

This is hosted on Peter Heinze's personal page on Hochschule Wismar - University of Technology, Business & Design.

Please remember to give credit to the aforementioned.

http://www.mb.hs-wismar.de/~heinze/subdir/FEA_of_Nonlinear_Problems_2011_Ivanco.pdf

Thanks a lot to everyone. You'll have been very helpful.

Hello, I need to assign ABS material properties in ANSYS. What information should I need to include in the material properties section? Should I assign nonlinear materials? If yes, which one should I select in ANSYS?

High deformations would be typical if the strain range is small, hence a non-linear analysis would be better suited.