Transgenic lines carrying a specific tissue tagged by green-fluorescence-protein (GFP) have been a powerful tool to developmental biology because they encapsulate the expression of endogenous genes. Traditionally with two-photon fluorescence microscopy based on a femtosecond Ti:sapphire laser (with a wavelength between 700-980nm), green fluorescence can be excited by simultaneous absorption of two photons for high-resolution three-dimensional (3D) optical imaging. However for in vivo biological applications, Ti:sapphire-laser based optical technology presents several limitations including finite penetration depth, strong on-focus cell damage, and phototoxicity. For high optical penetration and minimized photodamages, two-photon imaging based on light sources with an optical wavelength located around the biological penetration window (~1300nm) is desired, where unwanted light-tissue interactions including scattering, absorption, and photodamages can all be minimized. Previous experiments around the optical penetration window indicated inefficient green fluorescence excitation of GFP through three-photon absorption. Red fluorescence protein is thus highly desired for future non-invasive in vivo two-photon imaging. Screening from embryos injected with DNA fragment containing a heart-specific regulatory element of zebrafish cardiac myosin light chain 2 gene (cmlc2) fused with HcRed gene, we generate a zebrafish line that has strong two-photon red fluorescence expressed in cardiac cells based on a 1230nm femtosecond light source working in the biological penetration window. Combined with its nonlinearity, high penetration depth, and minimized photodamages, this method provides superb imaging capability compared with the traditional GFP based two-photon microscopy, offering deep insight into the noninvasive in vivo studies of gene expression in vertebrate embryos.