The submitted manuscript is a review of gas sensors based on various types of photonic crystal fibers (PCF), especially hollow-core PCFs. The review covers developments in this area that took place over a long period of time, since the first proposals of PCF fibers and their application to gas sensing, to the most recent works pursuing further advancement of properties of such sensors and proposing new sensing principles, by, for instance, specially tailoring the properties of PCF fibers.

Indeed, the concept of PCF fiber allows to control the light guiding properties in the most flexible ways, making it a perfect candidate for various sensing applications. In turn, hollow-core PCFs, as noted by the authors, allow to fill the core with whatever media is needed, which is extremely useful for gas and liquid sensing due to efficient interaction of light and the analyzed media while ensuring efficient light guiding and ability to perform remote, multiplexed and even distributed measurements.

The manuscript is well organized, the introduction put the reader in the context of gas sensing in general and stresses out advantages of optical fiber sensors application to this challenging task. The second section reviews general principles of gas sensing, from comparison to which, the advantages of HC-PCF fibers become obvious. The third and fourth sections give a good introduction into PCF fiber principles, which, in my opinion, is accessible even to those unfamiliar with such fibers. Various fiber types are described, the light guiding mechanisms of PCF are briefly discussed. The fifth and sixth sections describe various developed gas sensors based on PCF fibers, which sixth section emphasizing the advantages allowed by operating in MIR region. Finally, a short overview of the reviewed material is presented in the conclusion.

In my opinion, the manuscript presents a clear, unaltered view on PCF fibers application to gas sensing and will be of great importance to researches already working in the field as well as students and early-career researches and those who are less familiar with the topic. However, I have several suggestions, which, in my opinion, will help to further improve the quality and usefulness of this review:

Throughout the manuscript, the authors present the detection limits of various sensors they discuss, however, they use several different terms, such as sensitivity, noise equivalent concentration and detection limit. Typically, sensitivity of a sensor describes the relation between the physical quantity to be measured (gas concentration in terms of this manuscript) and and the quantity through which sensor interrogation is performed (intensity, wavelength, phase, etc.). Therefore, the term resolution or detection limit would be a better choice. Another point, concerning the detection limits is the bandwidth. When sensor resolution or detection limit is stated without the bandwidth in which the signal was detected, it is nearly meaningless and not suitable for comparison since in case of wider bandwidth the amount of noises is greater, hence, degrading the resolution. If, however, the signal is simply filtered, the detection bandwidth is decreased, therefore, reducing the noise level and improving the resolution, however, allowing for only slowly-varying signals to be detected. Therefore, my suggestion is to introduce resolution or detection limit normalized on the detection bandwidth (typically expressed in terms of units/(Hz^1/2)). Another suggestion is to elaborate a bit further on the problem of parasitic mode interference in side-hole PCFs and in case of a coupling gap between sensing PCF and feeding fiber. This will provide a better understanding of technological burdens and potential limitations of this type of sensors to the reader. Finally, I suggest to introduce a comparison table, in which various sensors discussed throughout the manuscript will be briefly compared in terms of sensing principle, operating spectral range, resolution, type of fiber with additional comments (N sensing elements multiplexed / distributed measurement with δL spatial resolution, etc.). Such tables can often be found in review papers and provide the reader with additional ability to quickly compare different principles/systems reported in the review. My overall suggestion is that the paper can be accepted after the changes are made according to the comments above.